Dataset Viewer
output
bool 2
classes | input
stringlengths 345
2.91k
|
|---|---|
true | Mechanistic Anomaly Detection for "Quirky'' Language Models
scalable oversight backdoor detection mechanistic interpretability outlier detection anomaly detection
As LLMs grow in capability, the task of supervising LLMs becomes more challenging. Supervision failures can occur if LLMs are sensitive to factors that supervisors are unaware of. We investigate __Mechanistic Anomaly Detection__ (MAD) as a technique to augment supervision of capable models; we use internal model features to identify anomalous training signals so they can be investigated or discarded. We train detectors to flag points from the test environment that differ substantially from the training environment, and experiment with a large variety of detector featuers and scoring rules to detect anomalies in a set of "quirky" language models. We find that detectors can achieve high discrimination on some tasks, but no detector is effective across all models and tasks. MAD techniques may be effective in low-stakes applications, but advances in both detection and evaluation are likely needed if they are to be used in high stakes settings. |
false | The Information-Autoencoding Family: A Lagrangian Perspective on Latent Variable Generative Modeling
Generative Models Variational Autoencoder Generative Adversarial Network
A variety of learning objectives have been recently proposed for training generative models. We show that many of them, including InfoGAN, ALI/BiGAN, ALICE, CycleGAN, VAE, $\beta$-VAE, adversarial autoencoders, AVB, and InfoVAE, are Lagrangian duals of the same primal optimization problem. This generalization reveals the implicit modeling trade-offs between flexibility and computational requirements being made by these models. Furthermore, we characterize the class of all objectives that can be optimized under certain computational constraints.
Finally, we show how this new Lagrangian perspective can explain undesirable behavior of existing methods and provide new principled solutions. |
false | Stochastic Proximal Point Algorithm for Large-scale Nonconvex Optimization: Convergence, Implementation, and Application to Neural Networks
sppa convergence nonconvex optimization implementation application sgd algorithm neural networks
We revisit the stochastic proximal point algorithm (SPPA) for large-scale nonconvex optimization problems. SPPA has been shown to converge faster and more stable than the celebrated stochastic gradient descent (SGD) algorithm, and its many variations, for convex problems. However, the per-iteration update of SPPA is defined abstractly and has long been considered expensive. In this paper, we show that efficient implementation of SPPA can be achieved. If the problem is a nonlinear least squares, each iteration of SPPA can be efficiently implemented by Gauss-Newton; with some linear algebra trick the resulting complexity is in the same order of SGD. For more generic problems, SPPA can still be implemented with L-BFGS or accelerated gradient with high efficiency. Another contribution of this work is the convergence of SPPA to a stationary point in expectation for nonconvex problems. The result is encouraging that it admits more flexible choices of the step sizes under similar assumptions. The proposed algorithm is elaborated for both regression and classification problems using different neural network structures. Real data experiments showcase its effectiveness in terms of convergence and accuracy compared to SGD and its variants. |
false | Hellinger Distance Constrained Regression
offline Reinforcement Learning off-policy control
This paper introduces an off-policy reinforcement learning method that uses Hellinger distance between sampling policy (from what samples were collected) and current policy (policy being optimized) as a constraint.
Hellinger distance squared multiplied by two is greater than or equal to total variation distance squared and less than or equal to Kullback-Leibler divergence, therefore a lower bound for expected discounted return for the new policy is improved compared to the lower bound for training with KL.
Also, Hellinger distance is less than or equal to 1, so there is a policy-independent lower bound for expected discounted return.
HDCR is capable of training with Experience Replay, a common setting for distributed RL when collecting trajectories using different policies and learning from this data centralized.
HDCR shows results comparable to or better than Advantage-weighted Behavior Model and Advantage-Weighted Regression on MuJoCo tasks using tiny offline datasets collected by random agents. On bigger datasets (100k timesteps) obtained by pretrained behavioral policy, HDCR outperforms ABM and AWR methods on 3 out of 4 tasks. |
true | Emergent Password Signalling in the Game of Werewolf
agents game werewolves emergent password signalling werewolf emergent communication efficient heuristics domain specificity better
Emergent communication can lead to more efficient problem-solving heuristics and more domain specificity. It can perform better than a handcrafted communication protocol, potentially directing autonomous agents towards unforeseen yet effective solutions. Previous research has investigated a social deduction game, called Werewolf, where two groups of autonomous agents, villagers and werewolves, interact in an environment named RLupus. We study the impact of allowing the agents to communicate through multiple rounds and evaluate their language and performance against the baseline environment. We show that agents develop a highly successful heuristic using a single word vocabulary. They create an approach using passwords, allowing them to determine which agents are werewolves, which is the winning condition. We explore the possible reasons behind this strategy, with further experimental analysis showing that our approach speeds up the convergence of the agents towards a common communication strategy. |
false | Unveiling Control Vectors in Language Models with Sparse Autoencoders
sparse autoencoder controllability model explanation mechanistic interpretability
Sparse autoencoders have recently emerged as a promising tool for explaining the internal mechanisms of large language models by disentangling complex activations into interpretable features. However, understanding the role and behavior of individual SAE features remains challenging. Prior approaches primarily focus on interpreting SAE features based on their activations or input correlations, which provide limited insight into their influence on model outputs. In this work, we investigate a specific subset of SAE features that directly control the generation behavior of LLMs. We term these “generation features”, as they reliably trigger the generation of specific tokens or semantically related token groups when activated, regardless of input context. Using a systematic methodology based on causal intervention, we identify and validate these features with significantly higher precision than baseline methods. Through extensive experiments on the Gemma models, we demonstrate that generation features reveal interesting phenomena about both the LLM and SAE architectures. These findings deepen our understanding of the generative mechanisms within LLMs and highlight the potential of SAEs for controlled text generation and model interpretability. Our code is available at https://anonymous.4open.science/r/control-vector-with-sae-AAFB. |
false | Truthfulness in LLMs: A Layer-wise Comparative Analysis of Representation Engineering and Contrast-Consistent Search
Truthfulness Large Language Models LLM Transparency Representation Engineering Contrast-Consistent Search Interpretability
The rapid advancement of Large Language Models (LLMs) has intensified the need for greater transparency in their internal representations. This study presents a layer-wise analysis of truthfulness storage in LLMs, comparing two state-of-the-art knowledge probing methodologies: Representation Engineering (RepE) and Contrast-Consistent Search (CCS). Our goal is to isolate truthfulness, defined as the factual accuracy of LLM outputs, from general knowledge encoded across model layers and to examine where and how this information is stored. RepE applies low-rank transformations within the model’s internal vector space, while CCS leverages pre-trained fixed vectors with an additional transformation layer to define truthfulness. Through experiments on Google’s Gemma models, evaluated across five diverse datasets, we find that truthfulness is embedded within pre-trained LLMs and can be amplified by specific input words. Our analysis reveals general trends in truthfulness storage and transferability, with CCS demonstrating greater stability in assessing truthfulness, while RepE exhibits potential in deeper layers but requires further refinement. Surprisingly, the truthfulness differences in the final layer, often considered the most critical, were statistically insignificant. This study provides empirical insights into the internal encoding of truthfulness in LLMs, highlighting the strengths and limitations of representation based transparency methods. |
true | Learning from others' mistakes: Avoiding dataset biases without modeling them
dataset bias product of experts natural language processing
State-of-the-art natural language processing (NLP) models often learn to model dataset biases and surface form correlations instead of features that target the intended underlying task. Previous work has demonstrated effective methods to circumvent these issues when knowledge of the bias is available. We consider cases where the bias issues may not be explicitly identified, and show a method for training models that learn to ignore these problematic correlations. Our approach relies on the observation that models with limited capacity primarily learn to exploit biases in the dataset. We can leverage the errors of such limited capacity models to train a more robust model in a product of experts, thus bypassing the need to hand-craft a biased model. We show the effectiveness of this method to retain improvements in out-of-distribution settings even if no particular bias is targeted by the biased model. |
true | PlasticineLab: A Soft-Body Manipulation Benchmark with Differentiable Physics
Soft Body Differentiable Physics Benchmark
Simulated virtual environments serve as one of the main driving forces behind developing and evaluating skill learning algorithms. However, existing environments typically only simulate rigid body physics. Additionally, the simulation process usually does not provide gradients that might be useful for planning and control optimizations. We introduce a new differentiable physics benchmark called PasticineLab, which includes a diverse collection of soft body manipulation tasks. In each task, the agent uses manipulators to deform the plasticine into a desired configuration. The underlying physics engine supports differentiable elastic and plastic deformation using the DiffTaichi system, posing many under-explored challenges to robotic agents. We evaluate several existing reinforcement learning (RL) methods and gradient-based methods on this benchmark. Experimental results suggest that 1) RL-based approaches struggle to solve most of the tasks efficiently; 2) gradient-based approaches, by optimizing open-loop control sequences with the built-in differentiable physics engine, can rapidly find a solution within tens of iterations, but still fall short on multi-stage tasks that require long-term planning. We expect that PlasticineLab will encourage the development of novel algorithms that combine differentiable physics and RL for more complex physics-based skill learning tasks. PlasticineLab will be made publicly available. |
false | PolyRetro: Few-shot Polymer Retrosynthesis via Domain Adaptation
ML for Chemistry Polymer Retrosynthesis Few-show Learning Domain Adaptation
Polymers appear everywhere in our daily lives -- fabrics, plastics, rubbers, etc. -- and we could hardly live without them. To make polymers, chemists develop processes that combine smaller building blocks~(monomers) to form long chains or complex networks~(polymers). These processes are called polymerizations and will usually take lots of human efforts to develop. Although machine learning models for small molecules have generated lots of promising results, the prediction problem for polymerization is new and suffers from the scarcity of polymerization datasets available in the field. Furthermore, the problem is made even more challenging by the large size of the polymers and the additional recursive constraints, which are not present in the small molecule problem. In this paper, we make an initial step towards this challenge and propose a learning-based search framework that can automatically identify a sequence of reactions that lead to the polymerization of a target polymer with minimal polymerization data involved. Our method transfers models trained on small molecule datasets for retrosynthesis to check the validity of polymerization reaction. Furthermore, our method also incorporates a template prior learned on a limited amount of polymer data into the framework to adapt the model from small molecule to the polymer domain. We demonstrate that our method is able to propose high-quality polymerization plans for a dataset of 52 real-world polymers, of which a significant portion successfully recovers the currently-in-used polymerization processes in the real world. |
false | EarlyBERT: Efficient BERT Training via Early-bird Lottery Tickets
Natural Language Processing Lottery Tickets Hypothesis Efficient Training
Deep, heavily overparameterized language models such as BERT, XLNet and T5 have achieved impressive success in many NLP tasks. However, their high model complexity requires enormous computation resources and extremely long training time for both pre-training and fine-tuning. Many works have studied model compression on large NLP models, but only focus on reducing inference cost/time, while still requiring expensive training process. Other works use extremely large batch sizes to shorten the pre-training time at the expense of high demand for computation resources. In this paper, inspired by the Early-Bird Lottery Tickets studied for computer vision tasks, we propose EarlyBERT, a general computationally-efficient training algorithm applicable to both pre-training and fine-tuning of large-scale language models. We are the first to identify structured winning tickets in the early stage of BERT training, and use them for efficient training. Comprehensive pre-training and fine-tuning experiments on GLUE and SQuAD downstream tasks show that EarlyBERT easily achieves comparable performance to standard BERT with 35~45% less training time. |
true | Unsupervised Functional Dependency Discovery for Data Preparation
Functional Dependencies Sparse Regression Structure Learning L1-regularization Weak Supervision
We study the problem of functional dependency (FD) discovery to impose domain knowledge for downstream data preparation tasks. We introduce a framework in which learning functional dependencies corresponds to solving a sparse regression problem. We show that our methods can scale to large data instances with millions of tuples and hundreds of attributes, while recovering true FDs across a diverse array of synthetic datasets, even in the presence of noisy data. Overall, our methods show an average F1 improvement of 2× against state-of-the-art FD discovery methods. Our system also obtains better F1 in downstream data repairing task than manually defined FDs.
|
false | Adaptive Gradient Methods Can Be Provably Faster than SGD with Random Shuffling
adaptive gradient methods sgd random many tasks neural networks acceleration effect setting best convergence rate worse literature
Adaptive gradient methods have been shown to outperform SGD in many tasks of training neural networks. However, the acceleration effect is yet to be explained in the non-convex setting since the best convergence rate of adaptive gradient methods is worse than that of SGD in literature. In this paper, we prove that adaptive gradient methods exhibit an $\small\tilde{O}(T^{-1/2})$-convergence rate for finding first-order stationary points under the strong growth condition, which improves previous best convergence results of adaptive gradient methods and random shuffling SGD by factors of $\small O(T^{-1/4})$ and $\small O(T^{-1/6})$, respectively. In particular, we study two variants of AdaGrad with random shuffling for finite sum minimization. Our analysis suggests that the combination of random shuffling and adaptive learning rates gives rise to better convergence. |
true | Neural Nonmyopic Bayesian Optimization in Dynamic Cost Settings
nonmyopic Bayesian optimization dynamic cost settings neural network policies gaussian process large language models
Bayesian optimization (BO) is a popular framework for optimizing black-box functions, leveraging probabilistic models such as Gaussian processes. Conventional BO algorithms, however, assume static query costs, which limit their applicability to real-world problems with dynamic cost structures such as geological surveys or biological sequence design, where query costs vary based on the previous actions. We propose a novel nonmyopic BO algorithm named LookaHES featuring dynamic cost models to address this. LookaHES employs a neural network policy for variational optimization over multi-step lookahead horizons to enable planning under dynamic cost environments. Empirically, we benchmark LookaHES on synthetic functions exhibiting varied dynamic cost structures. We subsequently apply LookaHES to a real-world application in protein sequence design using a large language model policy, demonstrating its scalability and effectiveness in handling multi-step planning in a large and complex query space. LookaHES consistently outperforms its myopic counterparts in synthetic and real-world settings, significantly improving efficiency and solution quality. Our implementation is available at https://github.com/sangttruong/nonmyopia. |
true | Identifying nonlinear dynamical systems with multiple time scales and long-range dependencies
nonlinear dynamical systems recurrent neural networks attractors computational neuroscience vanishing gradient problem LSTM
A main theoretical interest in biology and physics is to identify the nonlinear dynamical system (DS) that generated observed time series. Recurrent Neural Networks (RNN) are, in principle, powerful enough to approximate any underlying DS, but in their vanilla form suffer from the exploding vs. vanishing gradients problem. Previous attempts to alleviate this problem resulted either in more complicated, mathematically less tractable RNN architectures, or strongly limited the dynamical expressiveness of the RNN.
Here we address this issue by suggesting a simple regularization scheme for vanilla RNN with ReLU activation which enables them to solve long-range dependency problems and express slow time scales, while retaining a simple mathematical structure which makes their DS properties partly analytically accessible. We prove two theorems that establish a tight connection between the regularized RNN dynamics and their gradients, illustrate on DS benchmarks that our regularization approach strongly eases the reconstruction of DS which harbor widely differing time scales, and show that our method is also en par with other long-range architectures like LSTMs on several tasks. |
true | Can GANs Recover Faults in Electrical Motor Sensors?
GANs Electrical Motors Sensors
Electrical motors in industrial and emerging applications such as electrical automotive require high dynamic performance, robustness against parameter variation, and reliability. Recent advances in neural network-based estimators and fault detection techniques rely heavily on accurate sensor information. Due to the extreme operating conditions of electrical motors, there is always a chance of sensor failure which might lead to poor performance in downstream tasks using neural networks. This paper introduces the problem of identifying and recovering sensor faults using generative adversarial networks. We consider sensors monitoring various quantities like currents, voltages, speed, torque, temperature, and vibrations. We introduce fault model for these sensors to simulate training datasets. We use existing GAN based data imputation methods as baseline solutions.
Project Page: https://sagarverma.github.io/sensors.html |
true | Euclidean geometry meets graph, a geometric deep learning perspective
graph neural network geometric deep learning
Graph neural networks (GNN) have been an active area of machine learning research to tackle various problems in graph data. Sometimes nodes and edges in a graph can have spatial features, such as 3D coordinates of nodes and directions along edges. How do we reason over the topology of graphs while considering those geometric features? In this post, we discuss a ICLR 2021 paper "Learning from Protein Structure with Geometric Vector Perceptrons (GVP)". We take a geometric deep learning perspective to explain the intuition and advantage of GVP-GNN, the novel GNN architecture described in the paper, to reason over geometric graphs. We inspire an idea along physics-oriented interpretation. We also highlight under-appreciated applications of geometric graph learning in solving problems from multiple domains, including biochemistry, geography, and network analysis.
|
false | Deep Generative Neural Embeddings for High Dimensional Data Visualization
Deep Generative Models Embedding
We propose an embedding-based visualization method along with a data generation model. In particular, corresponding locations of data points in the visualization are optimized as embeddings along with a generative network such that the network could reconstruct the original data. The generalization aspect of the neural network enforces similar data points to be close in the embedding space. Since our method includes the generative part, it allows visualizations that are not possible with neighborhood embedding methods such as TSNE. Compared to parametric methods such as VAE our method is non-parametric and relaxes the need to optimize the encoder part, allowing us to obtain better optimizations. |
true | Improved Estimation of Concentration Under $\ell_p$-Norm Distance Metrics Using Half Spaces
Adversarial Examples Concentration of Measure Gaussian Isoperimetric Inequality
Concentration of measure has been argued to be the fundamental cause of adversarial vulnerability. Mahloujifar et al. (2019) presented an empirical way to measure the concentration of a data distribution using samples, and employed it to find lower bounds on intrinsic robustness for several benchmark datasets. However, it remains unclear whether these lower bounds are tight enough to provide a useful approximation for the intrinsic robustness of a dataset. To gain a deeper understanding of the concentration of measure phenomenon, we first extend the Gaussian Isoperimetric Inequality to non-spherical Gaussian measures and arbitrary $\ell_p$-norms ($p \geq 2$). We leverage these theoretical insights to design a method that uses half-spaces to estimate the concentration of any empirical dataset under $\ell_p$-norm distance metrics. Our proposed algorithm is more efficient than Mahloujifar et al. (2019)'s, and experiments on synthetic datasets and image benchmarks demonstrate that it is able to find much tighter intrinsic robustness bounds. These tighter estimates provide further evidence that rules out intrinsic dataset concentration as a possible explanation for the adversarial vulnerability of state-of-the-art classifiers. |
true | Unsupervised Demixing of Structured Signals from Their Superposition Using GANs
Demixing Generative Models GAN Unsupervised Learning Structured Recovery
Recently, Generative Adversarial Networks (GANs) have emerged as a popular alternative for modeling complex high dimensional distributions. Most of the existing works implicitly assume that the clean samples from the target distribution are easily available. However, in many applications, this assumption is violated. In this paper, we consider the observation setting in which the samples from a target distribution are given by the superposition of two structured components, and leverage GANs for learning of the structure of the components. We propose a novel framework, demixing-GAN, which learns the distribution of two components at the same time. Through extensive numerical experiments, we demonstrate that the proposed framework can generate clean samples from unknown distributions, which further can be used in demixing of the unseen test images. |
true | Learning Structural Edits via Incremental Tree Transformations
Tree-structured Data Edit Incremental Tree Transformations Representation Learning Imitation Learning Source Code
While most neural generative models generate outputs in a single pass, the human creative process is usually one of iterative building and refinement. Recent work has proposed models of editing processes, but these mostly focus on editing sequential data and/or only model a single editing pass. In this paper, we present a generic model for incremental editing of structured data (i.e. ''structural edits''). Particularly, we focus on tree-structured data, taking abstract syntax trees of computer programs as our canonical example. Our editor learns to iteratively generate tree edits (e.g. deleting or adding a subtree) and applies them to the partially edited data, thereby the entire editing process can be formulated as consecutive, incremental tree transformations. To show the unique benefits of modeling tree edits directly, we further propose a novel edit encoder for learning to represent edits, as well as an imitation learning method that allows the editor to be more robust. We evaluate our proposed editor on two source code edit datasets, where results show that, with the proposed edit encoder, our editor significantly improves accuracy over previous approaches that generate the edited program directly in one pass. Finally, we demonstrate that training our editor to imitate experts and correct its mistakes dynamically can further improve its performance. |
false | Learning to Use Future Information in Simultaneous Translation
sequence learning simultaneous machine translation
Simultaneous neural machine translation (briefly, NMT) has attracted much attention recently. In contrast to standard NMT, where the NMT system can access the full input sentence, simultaneous NMT is a prefix-to-prefix problem, where the system can only utilize the prefix of the input sentence and thus more uncertainty and difficulty are introduced to decoding. Wait-k inference is a simple yet effective strategy for simultaneous NMT, where the decoder generates the output sequence $k$ words behind the input words. For wait-k inference, we observe that wait-m training with $m>k$ in simultaneous NMT (i.e., using more future information for training than inference) generally outperforms wait-k training. Based on this observation, we propose a method that automatically learns how much future information to use in training for simultaneous NMT. Specifically, we introduce a controller to adaptively select wait-m training strategies according to the network status of the translation model and current training sentence pairs, and the controller is jointly trained with the translation model through bi-level optimization. Experiments on four datasets show that our method brings 1 to 3 BLEU point improvement over baselines under the same latency. Our code is available at https://github.com/P2F-research/simulNMT . |
true | Buy 4 REINFORCE Samples, Get a Baseline for Free!
reinforce multiple samples baseline sequence generation structured prediction travelling salesman problem
REINFORCE can be used to train models in structured prediction settings to directly optimize the test-time objective. However, the common case of sampling one prediction per datapoint (input) is data-inefficient. We show that by drawing multiple samples (predictions) per datapoint, we can learn with significantly less data, as we freely obtain a REINFORCE baseline to reduce variance. Additionally we derive a REINFORCE estimator with baseline, based on sampling without replacement. Combined with a recent technique to sample sequences without replacement using Stochastic Beam Search, this improves the training procedure for a sequence model that predicts the solution to the Travelling Salesman Problem. |
true | INFERNO: Inferring Object-Centric 3D Scene Representations without Supervision
object-centric representation learning nerf differentiable renderer autoencoder
We propose INFERNO, a method to infer object-centric representations of visual scenes without annotations.
Our method decomposes a scene into multiple objects, with each object having a structured representation that disentangles its shape, appearance and pose.
Each object representation defines a localized neural radiance field used to generate 2D views of the scene through differentiable rendering.
Our model is subsequently trained by minimizing a reconstruction loss between inputs and corresponding rendered scenes.
We empirically show that INFERNO discovers objects in a scene without supervision.
We also validate the interpretability of the learned representations by manipulating inferred scenes and showing the corresponding effect in the rendered output.
Finally, we demonstrate the usefulness of our 3D object representations in a visual reasoning task using the CATER dataset. |
false | A Chaos Theory Approach to Understand Neural Network Optimization
learning theory stochastic gradient descent deep learning neural networks dynamical systems chaos theory Lyapunov exponents
Despite the complicated structure of modern deep neural network architectures, they are still optimized with algorithms based on Stochastic Gradient Descent (SGD). However, the reason behind the effectiveness of SGD is not well understood, making its study an active research area. In this paper, we formulate deep neural network optimization as a dynamical system and show that the rigorous theory developed to study chaotic systems can be useful to understand SGD and its variants. In particular, we first observe that the inverse of the instability timescale of SGD optimization, represented by the largest Lyapunov exponent, corresponds to the most negative eigenvalue of the Hessian of the loss. This observation enables the introduction of an efficient method to estimate the largest eigenvalue of the Hessian. Then, we empirically show that for a large range of learning rates, SGD traverses the loss landscape across regions with largest eigenvalue of the Hessian similar to the inverse of the learning rate. This explains why effective learning rates can be found to be within a large range of values and shows that SGD implicitly uses the largest eigenvalue of the Hessian while traversing the loss landscape. This sheds some light on the effectiveness of SGD over more sophisticated second-order methods. We also propose a quasi-Newton method that dynamically estimates an optimal learning rate for the optimization of deep learning models. We demonstrate that our observations and methods are robust across different architectures and loss functions on CIFAR-10 dataset. |
false | The Advantage Regret-Matching Actor-Critic
Nash Equilibrium Games CFR
Regret minimization has played a key role in online learning, equilibrium computation in games, and reinforcement learning (RL). In this paper, we describe a general model-free RL method for no-regret learning based on repeated reconsideration of past behavior: Advantage Regret-Matching Actor-Critic (ARMAC). Rather than saving past state-action data, ARMAC saves a buffer of past policies, replaying through them to reconstruct hindsight assessments of past behavior. These retrospective value estimates are used to predict conditional advantages which, combined with regret matching, produces a new policy. In particular, ARMAC learns from sampled trajectories in a centralized training setting, without requiring the application of importance sampling commonly used in Monte Carlo counterfactual regret (CFR) minimization; hence, it does not suffer from excessive variance in large environments. In the single-agent setting, ARMAC shows an interesting form of exploration by keeping past policies intact. In the multiagent setting, ARMAC in self-play approaches Nash equilibria on some partially-observable zero-sum benchmarks. We provide exploitability estimates in the significantly larger game of betting-abstracted no-limit Texas Hold'em. |
false | Information Theoretic Meta Learning with Gaussian Processes
Meta Learning Information Bottleneck Gaussian Processes Few-shot learning Variational Inference
We formulate meta learning using information theoretic concepts such as mutual information and the information bottleneck. The idea is to learn a stochastic representation or encoding of the task description, given by a training or support set, that is highly informative about predicting the validation set. By making use of variational approximations to the mutual information, we derive a general and tractable framework for meta learning. We particularly develop new memory-based meta learning algorithms based on Gaussian processes and derive extensions that combine memory and gradient-based meta learning. We demonstrate our method on few-shot regression and classification by using standard benchmarks such as Omniglot, mini-Imagenet and Augmented Omniglot.
|
false | Sparse Uncertainty Representation in Deep Learning with Inducing Weights
Bayesian neural networks uncertainty estimation memory efficiency
Bayesian neural networks and deep ensembles represent two modern paradigms of uncertainty quantification in deep learning. Yet these approaches struggle to scale mainly due to memory inefficiency issues, since they require parameter storage several times higher than their deterministic counterparts. To address this, we augment the weight matrix of each layer with a small number of inducing weights, thereby projecting the uncertainty quantification into such low dimensional spaces. We further extend Matheron's conditional Gaussian sampling rule to enable fast weight sampling, whichenable our inference method to maintain reasonable run-time as compared with ensembles. Importantly, our approach achieves competitive performance to the state-of-the-art in prediction and uncertainty estimation tasks with fully connected neural networks and ResNets, while reducing the parameter size to $\leq 47.9\%$ of that of a single neural network. |
true | LiftPool: Bidirectional ConvNet Pooling
bidirectional pooling
Pooling is a critical operation in convolutional neural networks for increasing receptive fields and improving robustness to input variations. Most existing pooling operations downsample the feature maps, which is a lossy process. Moreover, they are not invertible: upsampling a downscaled feature map can not recover the lost information in the downsampling. By adopting the philosophy of the classical Lifting Scheme from signal processing, we propose LiftPool for bidirectional pooling layers, including LiftDownPool and LiftUpPool. LiftDownPool decomposes a feature map into various downsized sub-bands, each of which contains information with different frequencies. As the pooling function in LiftDownPool is perfectly invertible, by performing LiftDownPool backward, a corresponding up-pooling layer LiftUpPool is able to generate a refined upsampled feature map using the detail subbands, which is useful for image-to-image translation challenges. Experiments show the proposed methods achieve better results on image classification and semantic segmentation, using various backbones. Moreover, LiftDownPool offers better robustness to input corruptions and perturbations. |
true | UNLOCKING HIERARCHICAL CONCEPT DISCOVERY IN LANGUAGE MODELS THROUGH GEOMETRIC REGULARIZATION
Al Safety Trustworthy AI Hierarchical Representation Learning Mechanistic interpretability Sparse Autoencoder Feature Absorption Feature Splitting
We present Exponentially-Weighted Group Sparse Autoencoders (EWG-SAE) that aims to balance reconstruction quality and feature sparsity whilst resolving emerging problem such as feature absorption in interpretable language model analysis in a linguistically principled way through geometrically decaying group sparsity. Current sparse autoencoders struggle with merged hierarchical features due to uniform regularization encouraging absorption of broader features into more specific ones (e.g., "starts with S" being absorbed into "short"). Our architecture introduces hierarchical sparsity via $K=9$ dimension groups with exponential regularization decay ($\lambda_k = \lambda_{base} \times 0.5^k$), reducing absorption while maintaining state-of-the-art reconstruction fidelity, sparse probing score, and decent $\ell_1$ loss. The geometric structure enables precise feature isolation with negative inter-group correlations confirming hierarchical organization. |
true | Why Do Multiagent Systems Fail?
multi-agent systems large language models llm compound ai systems agents ai tool calling
Despite growing enthusiasm for Multi-Agent Systems (MAS), where multiple LLM agents collaborate to accomplish tasks, their performance gains across popular benchmarks remain minimal compared to single-agent frameworks. This gap highlights the need to analyze the challenges hindering MAS effectiveness. In this paper we conduct the first comprehensive study of challenges of MAS across 5 popular Multi-Agent Systems over 150+ tasks. We conduct an investigation with four expert human annotators studying the MAS execution traces, identifying 18 fine-grained failure modes, and propose a comprehensive failure taxonomy applicable across systems. We group these fine-grained failure modes into four key categories: (i) specification ambiguities and misalignment, (ii) organizational breakdowns, (iii) inter-agent conflict and coordination gaps, and (iv) weak verification and quality control. To understand whether these failure modes could have easily been avoided, we propose two interventions: improved agents roles specification and orchestration strategies. We find that identified failures require more involved solutions and we outline a roadmap for future research in this space. To contribute towards better development of MAS, we will open source our dataset, including the agent conversation traces and human annotations. |
true | WHAI: Weibull Hybrid Autoencoding Inference for Deep Topic Modeling
whai weibull hybrid inference inference network deep topic big corpora hierarchy scalable fast
To train an inference network jointly with a deep generative topic model, making it both scalable to big corpora and fast in out-of-sample prediction, we develop Weibull hybrid autoencoding inference (WHAI) for deep latent Dirichlet allocation, which infers posterior samples via a hybrid of stochastic-gradient MCMC and autoencoding variational Bayes. The generative network of WHAI has a hierarchy of gamma distributions, while the inference network of WHAI is a Weibull upward-downward variational autoencoder, which integrates a deterministic-upward deep neural network, and a stochastic-downward deep generative model based on a hierarchy of Weibull distributions. The Weibull distribution can be used to well approximate a gamma distribution with an analytic Kullback-Leibler divergence, and has a simple reparameterization via the uniform noise, which help efficiently compute the gradients of the evidence lower bound with respect to the parameters of the inference network. The effectiveness and efficiency of WHAI are illustrated with experiments on big corpora.
|
false | Sparse Binary Neural Networks
Binary Neural Networks Sparsity Deep Neural Network Compression
Quantized neural networks are gaining popularity thanks to their ability to solve complex tasks with comparable accuracy as full-precision Deep Neural Networks (DNNs), while also reducing computational power and storage requirements and increasing the processing speed. These properties make them an attractive alternative for the development and deployment of DNN-based applications in Internet-Of-Things (IoT) devices. Among quantized networks, Binary Neural Networks (BNNs) have reported the largest speed-up. However, they suffer from a fixed and limited compression factor that may result insufficient for certain devices with very limited resources. In this work, we propose Sparse Binary Neural Networks, a novel model and training scheme that allows to introduce sparsity in BNNs by using positive 0/1 binary weights, instead of the -1/+1 weights used by state-of-the-art binary networks. As a result, our method is able to achieve a high compression factor and reduces the number of operations and parameters at inference time. We study the properties of our method through experiments on linear and convolutional networks over MNIST and CIFAR-10 datasets. Experiments confirm that SBNNs can achieve high compression rates and good generalization, while further reducing the operations of BNNs, making it a viable option for deploying DNNs in very cheap and low-cost IoT devices and sensors. |
false | Residual Gated Graph ConvNets
graph neural networks ConvNets RNNs pattern matching semi-supervised clustering
Graph-structured data such as social networks, functional brain networks, gene regulatory networks, communications networks have brought the interest in generalizing deep learning techniques to graph domains. In this paper, we are interested to design neural networks for graphs with variable length in order to solve learning problems such as vertex classification, graph classification, graph regression, and graph generative tasks. Most existing works have focused on recurrent neural networks (RNNs) to learn meaningful representations of graphs, and more recently new convolutional neural networks (ConvNets) have been introduced. In this work, we want to compare rigorously these two fundamental families of architectures to solve graph learning tasks. We review existing graph RNN and ConvNet architectures, and propose natural extension of LSTM and ConvNet to graphs with arbitrary size. Then, we design a set of analytically controlled experiments on two basic graph problems, i.e. subgraph matching and graph clustering, to test the different architectures. Numerical results show that the proposed graph ConvNets are 3-17% more accurate and 1.5-4x faster than graph RNNs. Graph ConvNets are also 36% more accurate than variational (non-learning) techniques. Finally, the most effective graph ConvNet architecture uses gated edges and residuality. Residuality plays an essential role to learn multi-layer architectures as they provide a 10% gain of performance. |
false | Learning Chess Blindfolded
Chess Transformers Language Modeling World State
Transformer language models have made tremendous strides in natural language understanding. However, the complexity of natural language makes it challenging to ascertain how accurately these models are tracking the world state underlying the text. Motivated by this issue, we consider the task of language modeling for the game of chess. Unlike natural language, chess notations describe a simple, constrained, and deterministic domain. Moreover, we observe that chess notation itself allows for directly probing the world state, without requiring any additional probing-related machinery. Additionally, we have access to a vast number of chess games coupled with the exact state at every move, allowing us to measure the impact of various ways of including grounding during language model training. Overall, we find that with enough training data, transformer language models can learn to track pieces and predict legal moves when trained solely from move sequences. However, in adverse circumstances (small training sets or prediction following long move histories), providing access to board state information during training can yield consistent improvements. |
false | UTF: Undertrained Tokens as Fingerprints —— A Novel Approach to LLM Identification
Large Language Model fingerprint
Fingerprinting large language models (LLMs) is essential for verifying model ownership, ensuring authenticity, and preventing misuse. Traditional fingerprinting methods often require significant computational overhead or white-box verification access. In this paper, we introduce UTF, a novel and efficient approach to fingerprinting LLMs by leveraging under-trained tokens. Under-trained tokens are tokens that the model has not fully learned during its training phase. By utilizing these tokens, we perform supervised fine-tuning to embed specific input-output pairs into the model. This process allows the LLM to produce predetermined outputs when presented with certain inputs, effectively embedding a unique fingerprint.
Our method has minimal overhead and impact on model's performance, and does not require white-box access to target model's ownership identification. Compared to existing fingerprinting methods, UTF is also more effective and robust to fine-tuning and random guess. |
false | TOWARDS NATURAL ROBUSTNESS AGAINST ADVERSARIAL EXAMPLES
adversarial examples neural odes deep neural networks upper bound neural networks weaker upper bound natural robustness towards natural robustness
Recent studies have shown that deep neural networks are vulnerable to adversarial examples, but most of the methods proposed to defense adversarial examples cannot solve this problem fundamentally. In this paper, we theoretically prove that there is an upper bound for neural networks with identity mappings to constrain the error caused by adversarial noises. However, in actual computations, this kind of neural network no longer holds any upper bound and is therefore susceptible to adversarial examples. Following similar procedures, we explain why adversarial examples can fool other deep neural networks with skip connections. Furthermore, we demonstrate that a new family of deep neural networks called Neural ODEs (Chen et al., 2018) holds a weaker upper bound. This weaker upper bound prevents the amount of change in the result from being too large. Thus, Neural ODEs have natural robustness against adversarial examples. We evaluate the performance of Neural ODEs compared with ResNet under three white-box adversarial attacks (FGSM, PGD, DI2-FGSM) and one black-box adversarial attack (Boundary Attack). Finally, we show that the natural robustness of Neural ODEs is even better than the robustness of neural networks that are trained with adversarial training methods, such as TRADES and YOPO.
|
true | A Flexible Large Language Models Guardrail Development Methodology Applied to Off-Topic Prompt Detection
LLM Guardrails Synthetic Data
Large Language Models (LLMs) are prone to off-topic misuse, where users may prompt these models to perform tasks beyond their in- tended scope. Current guardrails, which often rely on curated examples or custom classifiers, suffer from high false-positive rates, limited adaptability, and the impracticality of requiring real-world data that isn’t available in pre-production. In this paper, we introduce a flexible, data-free guardrail development methodology that addresses these challenges. By thoroughly defining the problem space qualitatively and passing this to an LLM to generate diverse prompts, we construct a synthetic dataset to benchmark and train off-topic guardrails that outperform heuristic approaches. Addition- ally, by framing the task as classifying whether the user prompt is relevant with respect to the system prompt, our guardrails effectively generalize to other misuse categories, including jailbreak and harmful prompts. Lastly, we further contribute to the field by open- sourcing both the synthetic dataset and the off-topic guardrail models, providing valuable resources for developing guardrails in pre-production environments and supporting future research and development in LLM safety. |
true | Rethinking LLM Bias Probing Using Lessons from the Social Sciences
bias probing LLMs EcoLevels interdisciplinary social bias psychological theory
The proliferation of LLM bias probes introduces three significant challenges: (1) we lack principled criteria for choosing appropriate probes, (2) we lack a system for reconciling conflicting results across probes, and (3) we lack formal frameworks for reasoning about when (and why) probe results will generalize to real user behavior. We address these challenges by systematizing LLM social bias probing using actionable insights from social sciences. We then introduce EcoLevels – a framework that helps (a) determine appropriate bias probes, (b) reconcile conflicting findings across probes, and (c) generate predictions about bias generalization. Overall, we ground our analysis in social science research because many LLM probes are direct applications of human probes, and these fields have faced similar challenges when studying social bias in humans. Based on our work, we argue that the next frontier of LLM bias probing can (and should) benefit from decades of social science research. |
false | On the Importance of Looking at the Manifold
Topological Learning GNN VAE
Data rarely lies on uniquely Euclidean spaces. Even data typically represented in regular domains, such as images, can have a higher level of relational information, either between data samples or even relations within samples, e.g., how the objects in an image are linked. With this perspective our data points can be enriched by explicitly accounting for this connectivity and analyzing them as a graph. Herein, we analyze various approaches for unsupervised representation learning and investigate the importance of considering topological information and its impact when learning representations. We explore a spectrum of models, ranging from uniquely learning representations based on the isolated features of the nodes (focusing on Variational Autoencoders), to uniquely learning representations based on the topology (using node2vec) passing through models that integrate both node features and topological information in a hybrid fashion. For the latter we use Graph Neural Networks, precisely Deep Graph Infomax (DGI), and an extension of the typical formulation of the VAE where the topological structure is accounted for via an explicit regularization of the loss (Graph-Regularized VAEs, introduced in this work). To extensively investigate these methodologies, we consider a wide variety of data types: synthetic data point clouds, MNIST, citation networks, and chemical reactions. We show that each of the representations learned by these models may have critical importance for further downstream tasks, and that accounting for the topological features can greatly improve the modeling capabilities for certain problems. We further provide a framework to analyze these, and future models under different scenarios and types of data. |
false | Wide-minima Density Hypothesis and the Explore-Exploit Learning Rate Schedule
deep learning learning rate generalization
Several papers argue that wide minima generalize better than narrow minima. In this paper, through detailed experiments that not only corroborate the generalization properties of wide minima, we also provide empirical evidence for a new hypothesis that the density of wide minima is likely lower than the density of narrow minima. Further, motivated by this hypothesis, we design a novel explore-exploit learning rate schedule. On a variety of image and natural language datasets, compared to their original hand-tuned learning rate baselines, we show that our explore-exploit schedule can result in either up to 0.84% higher absolute accuracy using the original training budget or up to 57% reduced training time while achieving the original reported accuracy. For example, we achieve state-of-the-art (SOTA) accuracy for IWSLT'14 (DE-EN) and WMT'14 (DE-EN) datasets by just modifying the learning rate schedule of a high performing model. |
false | Masked Label Prediction: Unified Message Passing Model for Semi-Supervised Classification
Unified Message Passing Model Graph Neural Network Label Propagation Algorithm Semi-Supervised Classification.
Graph neural network (GNN) and label propagation algorithm (LPA) are both message passing algorithms, which have achieved superior performance in semi-supervised classification. GNN performs \emph{feature propagation} by a neural network to make predictions, while LPA uses \emph{label propagation} across graph adjacency matrix to get results. However, there is still no good way to combine these two kinds of algorithms. In this paper, we proposed a new {\bf Uni}fied {\bf M}essage {\bf P}assaging Model (UniMP) that can incorporate \emph{feature propagation} and \emph{label propagation} with a shared message passing network, providing a better performance in semi-supervised classification. First, we adopt a Graph Transformer jointly label embedding to propagate both the feature and label information. Second, to train UniMP without overfitting in self-loop label information, we propose a masked label prediction strategy, in which some percentage of training labels are simply masked at random, and then predicted. UniMP conceptually unifies feature propagation and label propagation and be empirically powerful. It obtains new state-of-the-art semi-supervised classification results in Open Graph Benchmark (OGB). |
true | How Much Over-parameterization Is Sufficient to Learn Deep ReLU Networks?
deep ReLU networks neural tangent kernel (stochastic) gradient descent generalization error classification
A recent line of research on deep learning focuses on the extremely over-parameterized setting, and shows that when the network width is larger than a high degree polynomial of the training sample size $n$ and the inverse of the target error $\epsilon^{-1}$, deep neural networks learned by (stochastic) gradient descent enjoy nice optimization and generalization guarantees. Very recently, it is shown that under certain margin assumptions on the training data, a polylogarithmic width condition suffices for two-layer ReLU networks to converge and generalize (Ji and Telgarsky, 2020). However, whether deep neural networks can be learned with such a mild over-parameterization is still an open question. In this work, we answer this question affirmatively and establish sharper learning guarantees for deep ReLU networks trained by (stochastic) gradient descent. In specific, under certain assumptions made in previous work, our optimization and generalization guarantees hold with network width polylogarithmic in $n$ and $\epsilon^{-1}$. Our results push the study of over-parameterized deep neural networks towards more practical settings. |
false | SkillBERT: “Skilling” the BERT to classify skills!
skillbert skills bert features errs skills present experiments algorithms random forest age
In the age of digital recruitment, job posts can attract a large number of applications, and screening them manually can become a very tedious task. These recruitment records are stored in the form of tables in our recruitment database (Electronic Recruitment Records, referred to as ERRs). We have released a de-identified ERR dataset to the public domain. We also propose a BERT-based model, SkillBERT, the embeddings of which are used as features for classifying skills present in the ERRs into groups referred to as "competency groups". A competency group is a group of similar skills and it is used as matching criteria (instead of matching on skills) for finding the overlap of skills between the candidates and the jobs. This proxy match takes advantage of the BERT's capability of deriving meaning from the structure of competency groups present in the skill dataset. In our experiments, the SkillBERT, which is trained from scratch on the skills present in job requisitions, is shown to be better performing than the pre-trained BERT and the Word2Vec. We have also explored K-means clustering and spectral clustering on SkillBERT embeddings to generate cluster-based features. Both algorithms provide similar performance benefits. Last, we have experimented with different machine learning algorithms like Random Forest, XGBoost, and a deep learning algorithm Bi-LSTM . We did not observe a significant performance difference among the algorithms, although XGBoost and Bi-LSTM perform slightly better than Random Forest. The features created using SkillBERT are most predictive in the classification task, which demonstrates that the SkillBERT is able to capture information about the skills' ontology from the data. We have made the source code and the trained models of our experiments publicly available. |
false | On the Importance of Sampling in Training GCNs: Convergence Analysis and Variance Reduction
Graph neural network large-scale machine learning convergence analysis
Graph Convolutional Networks (GCNs) have achieved impressive empirical advancement across a wide variety of graph-related applications. Despite their great success, training GCNs on large graphs suffers from computational and memory issues. A potential path to circumvent these obstacles is sampling-based methods, where at each layer a subset of nodes is sampled. Although recent studies have empirically demonstrated the effectiveness of sampling-based methods, these works lack theoretical convergence guarantees under realistic settings and cannot fully leverage the information of evolving parameters during optimization. In this paper, we describe and analyze a general \textbf{\textit{doubly variance reduction}} schema that can accelerate any sampling method under the memory budget. The motivating impetus for the proposed schema is a careful analysis for the variance of sampling methods where it is shown that the induced variance can be decomposed into node embedding approximation variance (\emph{zeroth-order variance}) during forward propagation and layerwise-gradient variance (\emph{first-order variance}) during backward propagation. We theoretically analyze the convergence of the proposed schema and show that it enjoys an $\mathcal{O}(1/T)$ convergence rate. We complement our theoretical results by integrating the proposed schema in different sampling methods and applying them to different large real-world graphs. |
true | Analyzing Memorization in Large Language Models through the Lens of Model Attribution
Memorization Generalization Language Models Model Attribution Attention
Large Language Models (LLMs) are prevalent in modern applications but often memorize training data, leading to privacy breaches and copyright issues. Existing research has mainly focused on post-hoc analyses—such as extracting memorized content or developing memorization metrics—without exploring the underlying architectural factors contributing to memorization. In this work, we investigate memorization from an architectural lens by analyzing how attention modules at different layers impact its memorization and generalization performance. Using attribution techniques, we systematically intervene in the LLM's architecture by bypassing attention modules at specific blocks while keeping other components like layer normalization and MLP transformations intact. We provide theorems analyzing our intervention mechanism from a mathematical view, bounding the difference in layer outputs with and without our attributions. Our theoretical and empirical analyses reveal that attention modules in deeper transformer blocks are primarily responsible for memorization, whereas earlier blocks are crucial for the model's generalization and reasoning capabilities. We validate our findings through comprehensive experiments on different LLM families (Pythia and GPT-Neo) and five benchmark datasets. Our insights offer a practical approach to mitigate memorization in LLMs while preserving their performance, contributing to safer and more ethical deployment in real-world applications. |
true | Zebra: a continuous generative transformer for solving parametric PDEs
PDE Deep Learning Foundation Model Transformer
Foundation models have revolutionized deep learning, moving beyond task-specific architectures to versatile models pre-trained using self-supervised learning on extensive datasets. These models have set new benchmarks across domains, including natural language processing, computer vision, and biology, due to their adaptability and state-of-the-art performance on downstream tasks. Yet, for solving PDEs or modeling physical dynamics, the potential of foundation models remains untapped due to the limited scale of existing datasets. This study presents Zebra, a novel generative model that adapts language model techniques to the continuous domain of PDE solutions. Pre-trained on specific PDE families, Zebra excels in dynamic forecasting, surpassing existing neural operators and solvers, and establishes a promising path for foundation models extensively pre-trained on varied PDE scenarios to tackle PDE challenges with scarce data. |
true | Evaluation of Similarity-based Explanations
Interpretability Explainability
Explaining the predictions made by complex machine learning models helps users to understand and accept the predicted outputs with confidence. One promising way is to use similarity-based explanation that provides similar instances as evidence to support model predictions. Several relevance metrics are used for this purpose. In this study, we investigated relevance metrics that can provide reasonable explanations to users. Specifically, we adopted three tests to evaluate whether the relevance metrics satisfy the minimal requirements for similarity-based explanation. Our experiments revealed that the cosine similarity of the gradients of the loss performs best, which would be a recommended choice in practice. In addition, we showed that some metrics perform poorly in our tests and analyzed the reasons of their failure. We expect our insights to help practitioners in selecting appropriate relevance metrics and also aid further researches for designing better relevance metrics for explanations. |
false | SBEVNet: End-to-End Deep Stereo Layout Estimation
Layout Estimation Deep Stereo Computer Vision
Accurate layout estimation is crucial for planning and navigation, for robotics applications such as self driving. In this paper, we introduce stereo bird's eye view network SBEVNet, a novel supervised end-to-end framework for estimation of bird's eye view layout from a pair of stereo images. Although our network reuses the building blocks from the state-of-the-art deep learning networks for disparity estimation, we show that accurate depth estimation is neither sufficient nor necessary. Instead, the learning of a good internal bird's eye view feature representation is essential for layout estimation. Specifically, we first generate a disparity feature volume using the features of the stereo images and then project it to the bird's eye view coordinates. This gives us coarse grained scene structural information. We also apply inverse perspective mapping (IPM) to map the input images and their features to the bird's eye view. This gives us fine grained texture information. The concatenated IPM features with the projected feature volume creates a rich bird's eye view representation which is capable of spatial reasoning. We use this representation to estimate the BEV semantic map. Additionally, we show that using the IPM features as a supervisory signal for stereo features can give an improvement in performance. We demonstrate our approach on two datasets: KITTI dataset and synthetically generated dataset using the CARLA simulator. For both of the datasets, we establish state-of-the-art performance beyond other baselines. |
true | ExpProof : Operationalizing Explanations for Confidential Models with ZKPs
Explanation Trust LIME Zero-Knowledge Proofs Auditing
In principle, explanations are intended as a way to increase trust in machine learn-
ing models and are often obligated by regulations. However, many circumstances
where these are demanded are adversarial in nature, meaning the involved parties
have misaligned interests and are incentivized to manipulate explanations for their
purpose. As a result, explainability methods fail to be operational in such settings
despite the demand Bordt et al. (2022). In this paper, we take a step towards op-
erationalizing explanations in adversarial scenarios with Zero-Knowledge Proofs
(ZKPs), a cryptographic primitive. Specifically we explore ZKP-amenable ver-
sions of the popular explainability algorithm LIME and evaluate their performance
on Neural Networks and Random Forests. Our code is publicly available at : https://github.com/infinite-pursuits/ExpProof. |
true | Learning to Make Decisions via Submodular Regularization
decisions actions action submodular regularization tasks combinatorial optimization problems large number cost high
Many sequential decision making tasks can be viewed as combinatorial optimization problems over a large number of actions. When the cost of evaluating an action is high, even a greedy algorithm, which iteratively picks the best action given the history, is prohibitive to run. In this paper, we aim to learn a greedy heuristic for sequentially selecting actions as a surrogate for invoking the expensive oracle when evaluating an action. In particular, we focus on a class of combinatorial problems that can be solved via submodular maximization (either directly on the objective function or via submodular surrogates). We introduce a data-driven optimization framework based on the submodular-norm loss, a novel loss function that encourages the resulting objective to exhibit diminishing returns. Our framework outputs a surrogate objective that is efficient to train, approximately submodular, and can be made permutation-invariant. The latter two properties allow us to prove strong approximation guarantees for the learned greedy heuristic. Furthermore, we show that our model can be easily integrated with modern deep imitation learning pipelines for sequential prediction tasks. We demonstrate the performance of our algorithm on a variety of batched and sequential optimization tasks, including set cover, active learning, and Bayesian optimization for protein engineering. |
false | DocImpact: Quantifying Document Impact in RAG-LLMs
RAG Augmented Generation Document Influence LLM
We present DocImpact, a novel methodology for measuring the influence of individual documents in Retrieval-Augmented Generation (RAG) systems. While RAG architectures have become increasingly popular in modern language models, understanding the precise contribution of each retrieved document to model outputs remains challenging. Our algorithm employs a counterfactual analysis by systematically excluding individual documents and measuring the divergence in model outputs compared to the full-context baseline. We implement our RAG-LLM using Pinecone as the database and Llama-3.1-70b as the LLM. |
false | Linear Representation Meta-Reinforcement Learning for Instant Adaptation
meta reinforcement learning out-of-distribution reinforcement learning
This paper introduces Fast Linearized Adaptive Policy (FLAP), a new meta-reinforcement learning (meta-RL) method that is able to extrapolate well to out-of-distribution tasks without the need to reuse data from training, and adapt almost instantaneously with the need of only a few samples during testing. FLAP builds upon the idea of learning a shared linear representation of the policy so that when adapting to a new task, it suffices to predict a set of linear weights. A separate adapter network is trained simultaneously with the policy such that during adaptation, we can directly use the adapter network to predict these linear weights instead of updating a meta-policy via gradient descent such as in prior Meta-RL algorithms like MAML to obtain the new policy. The application of the separate feed-forward network not only speeds up the adaptation run-time significantly, but also generalizes extremely well to very different tasks that prior Meta-RL methods fail to generalize to. Experiments on standard continuous-control meta-RL benchmarks show FLAP presenting significantly stronger performance on out-of-distribution tasks with up to double the average return and up to 8X faster adaptation run-time speeds when compared to prior methods. |
false | GAN2Shape - Create 3D Shape with 2D GANs
3D Deep Learning GANs
In this blog post, we discuss the key points of the paper “Do 2D GANs Know 3D Shape? Unsupervised 3D Shape Reconstruction from 2D Image GANs” (GAN2Shape) by Pan et al. We will discuss both the theory and code (in the author's GitHub repository and use a demo Colab notebook to show how GAN2Shape is able to transform 2D images to 3D images in multiple view images format.
We gave a brief introduction to 3D deep learning, GANs, StyleGAN2, and Unsup3D . Then we discussed in detail the key steps of GAN2Shape: how to transform a 2D image into 3D shapes, by using Unsup3D and StyleGAN2D. We demonstrated the training process with a Colab notebook to show the input image and the generated 3D images. |
false | DISTRIBUTIONALLY ROBUST NEURAL NETWORKS FOR GROUP SHIFTS ON THE IMPORTANCE OF REGULARIZATION FOR WORST-CASE GENERALIZATION
generalization regularization Robustness Group shift
Overparameterized neural networks can be highly accurate on average on an i.i.d. test set yet consistently fail on atypical groups of the data (e.g., by learning spurious correlations that hold on average but not in such groups). Distributionally robust optimization (DRO) allows us to learn models that instead minimize the worst-case training loss over a set of pre-defined groups. However, we find that naively applying group DRO to overparameterized neural networks fails: these models can perfectly fit the training data, and any model with vanishing average training loss also already has vanishing worst-case training loss. Instead, the poor worst-case performance arises from poor generalization on some groups. By coupling group DRO models with increased regularization—a stronger-than-typical
$l_{2}$ penalty or early stopping—we achieve substantially higher worst-group accuracies, with 10–40 percentage point improvements on a natural language inference task and two image tasks, while maintaining high average accuracies. Our results suggest that regularization is important for worst-group generalization in the overparameterized regime, even if it is not needed for average generalization. Finally, we introduce a stochastic optimization algorithm, with convergence guarantees, to efficiently train group DRO models. |
true | Generalized Multimodal ELBO
Multimodal VAE ELBO self-supervised generative learning
Multiple data types naturally co-occur when describing real-world phenomena and learning from them is a long-standing goal in machine learning research. However, existing self-supervised generative models approximating an ELBO are not able to fulfill all desired requirements of multimodal models: their posterior approximation functions lead to a trade-off between the semantic coherence and the ability to learn the joint data distribution. We propose a new, generalized ELBO formulation for multimodal data that overcomes these limitations. The new objective encompasses two previous methods as special cases and combines their benefits without compromises. In extensive experiments, we demonstrate the advantage of the proposed method compared to state-of-the-art models in self-supervised, generative learning tasks. |
true | ResNet After All: Neural ODEs and Their Numerical Solution
ode training neural odes flow equal solver resnet numerical solution resnet numerical solution key appeal
A key appeal of the recently proposed Neural Ordinary Differential Equation (ODE) framework is that it seems to provide a continuous-time extension of discrete residual neural networks.
As we show herein, though, trained Neural ODE models actually depend on the specific numerical method used during training.
If the trained model is supposed to be a flow generated from an ODE, it should be possible to choose another numerical solver with equal or smaller numerical error without loss of performance.
We observe that if training relies on a solver with overly coarse discretization, then testing with another solver of equal or smaller numerical error results in a sharp drop in accuracy.
In such cases, the combination of vector field and numerical method cannot be interpreted as a flow generated from an ODE, which arguably poses a fatal breakdown of the Neural ODE concept.
We observe, however, that there exists a critical step size beyond which the training yields a valid ODE vector field.
We propose a method that monitors the behavior of the ODE solver during training to adapt its step size, aiming to ensure a valid ODE without unnecessarily increasing computational cost.
We verify this adaption algorithm on a common bench mark dataset as well as a synthetic dataset.
|
false | Design For Trustworthy AI Solutions
Trustworthy AI Data Interpretation Conformity Assessment Decision Intelligence Bias Parallel Data Imbalance Data Risk Assessment Sensitivity Loss Algorithmic Assessment Concept Test Adversarial AI What-if-Analysis Counterfactuals Explainability
Transparency of an AI solution is the need of the hour. With growing adoption, AI is increasingly making business critical decisions in organizations and propagating it, not only limited to organizations but also to society. This has resulted in growing legislative asks from organizations such as "US Algorithmic Accountability Act -2019", " EU Right To Explainability", "EU AI act 2021" and so on. We hereby propose a framework for addressing transparency in AI solution and bringing about trustworthiness, reliability and un-biasness of AI solutions to various stakeholders, which may include but not limited to, AI Solution Engineers, Chief Legal Council, Decision Reviewers etc. Our solution addresses the problem via providing transparency in terms of Data Interpretation, where we use AI to spot historical bias, mitigate them and perform risk assessment of the data; Conformity assessment, where we test trustworthiness, robustness and explainability of AI algorithm; Decision Intelligence, where we provide insights on financial impact, potential risks and scope for human intervention based on business and regulatory requirements. |
true | NOVAS: Non-convex Optimization via Adaptive Stochastic Search for End-to-end Learning and Control
deep neural networks nested optimization stochastic control deep FBSDEs
In this work we propose the use of adaptive stochastic search as a building block for general, non-convex optimization operations within deep neural network architectures. Specifically, for an objective function located at some layer in the network and parameterized by some network parameters, we employ adaptive stochastic search to perform optimization over its output. This operation is differentiable and does not obstruct the passing of gradients during backpropagation, thus enabling us to incorporate it as a component in end-to-end learning. We study the proposed optimization module's properties and benchmark it against two existing alternatives on a synthetic energy-based structured prediction task, and further showcase its use in stochastic optimal control applications. |
true | Storyboarding of Recipes: Grounded Contextual Generation
Multimodal Generation Structure Scaffold Multitask
Information need of humans is essentially multimodal in nature, enabling maximum exploitation of situated context. We introduce a dataset for sequential procedural (how-to) text generation from images in cooking domain. The dataset consists of 16,441 cooking recipes with 160,479 photos associated with different steps. We setup a baseline motivated by the best performing model in terms of human evaluation for the Visual Story Telling (ViST) task. In addition, we introduce two models to incorporate high level structure learnt by a Finite State Machine (FSM) in neural sequential generation process by: (1) Scaffolding Structure in Decoder (SSiD) (2) Scaffolding Structure in Loss (SSiL). These models show an improvement in empirical as well as human evaluation. Our best performing model (SSiL) achieves a METEOR score of 0.31, which is an improvement of 0.6 over the baseline model. We also conducted human evaluation of the generated grounded recipes, which reveal that 61% found that our proposed (SSiL) model is better than the baseline model in terms of overall recipes, and 72.5% preferred our model in terms of coherence and structure. We also discuss analysis of the output highlighting key important NLP issues for prospective directions.
|
true | What are the Statistical Limits of Offline RL with Linear Function Approximation?
batch reinforcement learning function approximation lower bound representation
Offline reinforcement learning seeks to utilize offline (observational) data to guide the learning of (causal) sequential decision making strategies. The hope is that offline reinforcement learning coupled with function approximation methods (to deal with the curse of dimensionality) can provide a means to help alleviate the excessive sample complexity burden in modern sequential decision making problems. However, the extent to which this broader approach can be effective is not well understood, where the literature largely consists of sufficient conditions.
This work focuses on the basic question of what are necessary representational and distributional conditions that permit provable sample-efficient offline reinforcement learning. Perhaps surprisingly, our main result shows that even if: i) we have realizability in that the true value function of \emph{every} policy is linear in a given set of features and 2) our off-policy data has good coverage over all features (under a strong spectral condition), any algorithm still (information-theoretically) requires a number of offline samples that is exponential in the problem horizon to non-trivially estimate the value of \emph{any} given policy. Our results highlight that sample-efficient offline policy evaluation is not possible unless significantly stronger conditions hold; such conditions include either having low distribution shift (where the offline data distribution is close to the distribution of the policy to be evaluated) or significantly stronger representational conditions (beyond realizability). |
true | Online Semi-Supervised Learning with Bandit Feedback
online learning graph convolutional networks contextual bandits
We formulate a new problem at the intersection of semi-supervised learning and contextual bandits, motivated by several applications including clinical trials and dialog systems. We demonstrate how contextual bandit and graph convolutional networks can be adjusted to the new problem formulation. We then take the best of both approaches to develop multi-GCN embedded contextual bandit. Our algorithms are verified on several real world datasets. |
true | TUCKER DECOMPOSITION FOR INTERPRETABLE NEURAL ORDINARY DIFFERENTIAL EQUATIONS
Neural ODEs Interpretability Tensor decomposition dynamical systems Polynomial Networks
Tensorial and polynomial networks have emerged as effective tools in vari- ous fields, particularly for modeling the multilinear relationships among input variables. More recently, polynomial networks factorized using the canonical polyadic tensor decomposition (CPD) have been successfully used in the prob- lem of system dynamics identification, where the relations between variables are usually a polynomial function. This paper introduces a more general tensorial net- work that employs Tucker decomposition, thereby providing enhanced flexibility and expressivity in model construction. The study evaluates the performance of TuckerNet, comparing it against CPD-based networks in learning functions and identifying ordinary differential equation dynamics. The findings demonstrate the potential of TuckerNet as a superior alternative for tensorial network construction, particularly when constraining the number of parameters, while also highlighting aspects beyond decomposition that impact learning outcomes. |
true | Neural Networks for Learning Counterfactual G-Invariances from Single Environments
Extrapolation G-invariance regularization Counterfactual inference Invariant subspaces
Despite —or maybe because of— their astonishing capacity to fit data, neural networks are believed to have difficulties extrapolating beyond training data distribution. This work shows that, for extrapolations based on finite transformation groups, a model’s inability to extrapolate is unrelated to its capacity. Rather, the shortcoming is inherited from a learning hypothesis: Examples not explicitly observed with infinitely many training examples have underspecified outcomes in the learner’s model. In order to endow neural networks with the ability to extrapolate over group transformations, we introduce a learning framework counterfactually-guided by the learning hypothesis that any group invariance to (known) transformation groups is mandatory even without evidence, unless the learner deems it inconsistent with the training data. Unlike existing invariance-driven methods for (counterfactual) extrapolations, this framework allows extrapolations from a single environment. Finally, we introduce sequence and image extrapolation tasks that validate our framework and showcase the shortcomings of traditional approaches. |
false | AUL is a better optimization metric in PU learning
data aul better optimization metric model performance pu learning aul common real life dataset
Traditional binary classification models are trained and evaluated with fully labeled data which is not common in real life. In non-ideal dataset, only a small fraction of positive data are labeled. Training a model from such partially labeled data is named as positive-unlabeled (PU) learning. A naive solution of PU learning is treating unlabeled samples as negative. However, using biased data, the trained model may converge to non-optimal point and its real performance cannot be well estimated. Recent works try to recover the unbiased result by estimating the proportion of positive samples with mixture proportion estimation (MPE) algorithms, but the model performance is still limited and heavy computational cost is introduced (particularly for big datasets). In this work, we theoretically prove that Area Under Lift curve (AUL) is an unbiased metric in PU learning scenario, and the experimental evaluation on 9 datasets shows that the average absolute error of AUL estimation is only 1/6 of AUC estimation. By experiments we also find that, compared with state-of-the-art AUC-optimization algorithm, AULoptimization algorithm can not only significantly save the computational cost, but also improve the model performance by up to 10%. |
false | Disentangled Representation Learning with Information Maximizing Autoencoder
Disentangled Representation Learning Data Augmentation Generative Adversarial Nets Unsupervised Learning
Learning disentangled representation from any unlabelled data is a non-trivial problem. In this paper we propose Information Maximising Autoencoder (InfoAE) where the encoder learns powerful disentangled representation through maximizing the mutual information between the representation and given information in an unsupervised fashion. We have evaluated our model on MNIST dataset and achieved approximately 98.9 % test accuracy while using complete unsupervised training. |
false | Learning to Optimize Neural Nets
Learning to learn meta-learning reinforcement learning optimization
Learning to Optimize is a recently proposed framework for learning optimization algorithms using reinforcement learning. In this paper, we explore learning an optimization algorithm for training shallow neural nets. Such high-dimensional stochastic optimization problems present interesting challenges for existing reinforcement learning algorithms. We develop an extension that is suited to learning optimization algorithms in this setting and demonstrate that the learned optimization algorithm consistently outperforms other known optimization algorithms even on unseen tasks and is robust to changes in stochasticity of gradients and the neural net architecture. More specifically, we show that an optimization algorithm trained with the proposed method on the problem of training a neural net on MNIST generalizes to the problems of training neural nets on the Toronto Faces Dataset, CIFAR-10 and CIFAR-100. |
true | HaluEval-Wild: Evaluating Hallucinations of Language Models in the Wild
Large Language Models Hallucination Factuality Trustworthy Evaluation
Hallucinations pose a significant challenge to the reliability of large language models (LLMs) in critical domains. Recent benchmarks designed to assess LLM hallucinations within conventional NLP tasks, such as knowledge-intensive question answering (QA) and summarization, are insufficient for capturing the complexities of user-LLM interactions in dynamic, real-world settings. To address this gap, we introduce HaluEval-Wild, the first benchmark specifically designed to evaluate LLM hallucinations in the wild. We meticulously collect challenging (adversarially filtered by Alpaca) user queries from ShareGPT, an existing real-world user-LLM interaction datasets, to evaluate the hallucination rates of various LLMs. Upon analyzing the collected queries, we categorize them into five distinct types, which enables a fine-grained analysis of the types of hallucinations LLMs exhibit, and synthesize the reference answers with the powerful GPT-4 model and retrieval-augmented generation (RAG). Our benchmark offers a novel approach towards enhancing our comprehension of and improving LLM reliability in scenarios reflective of real-world interactions. |
true | Group Equivariant Stand-Alone Self-Attention For Vision
group equivariant transformers group equivariant self-attention group equivariance self-attention transformers
We provide a general self-attention formulation to impose group equivariance to arbitrary symmetry groups. This is achieved by defining positional encodings that are invariant to the action of the group considered. Since the group acts on the positional encoding directly, group equivariant self-attention networks (GSA-Nets) are steerable by nature. Our experiments on vision benchmarks demonstrate consistent improvements of GSA-Nets over non-equivariant self-attention networks. |
true | HYPE: Human-eYe Perceptual Evaluation of Generative Models
Generative models generative evaluation evaluation metric human evaluation
Generative models often use human evaluations to determine and justify progress. Unfortunately, existing human evaluation methods are ad-hoc: there is currently no standardized, validated evaluation that: (1) measures perceptual fidelity, (2) is reliable, (3) separates models into clear rank order, and (4) ensures high-quality measurement without intractable cost. In response, we construct Human-eYe Perceptual Evaluation (HYPE), a human metric that is (1) grounded in psychophysics research in perception, (2) reliable across different sets of randomly sampled outputs from a model, (3) results in separable model performances, and (4) efficient in cost and time. We introduce two methods. The first, HYPE-Time, measures visual perception under adaptive time constraints to determine the minimum length of time (e.g., 250ms) that model output such as a generated face needs to be visible for people to distinguish it as real or fake. The second, HYPE-Infinity, measures human error rate on fake and real images with no time constraints, maintaining stability and drastically reducing time and cost. We test HYPE across four state-of-the-art generative adversarial networks (GANs) on unconditional image generation using two datasets, the popular CelebA and the newer higher-resolution FFHQ, and two sampling techniques of model outputs. By simulating HYPE's evaluation multiple times, we demonstrate consistent ranking of different models, identifying StyleGAN with truncation trick sampling (27.6% HYPE-Infinity deception rate, with roughly one quarter of images being misclassified by humans) as superior to StyleGAN without truncation (19.0%) on FFHQ. |
false | Interpretability Through Invertibility: A Deep Convolutional Network With Ideal Counterfactuals And Isosurfaces
Interpretable Machine Learning Counterfactuals Computer Vision Human Evaluation User Study
Current state of the art computer vision applications rely on highly complex models. Their interpretability is mostly limited to post-hoc methods which are not guaranteed to be faithful to the model. To elucidate a model’s decision, we present a novel interpretable model based on an invertible deep convolutional network. Our model generates meaningful, faithful, and ideal counterfactuals. Using PCA on the classifier’s input, we can also create “isofactuals”– image interpolations with the same outcome but visually meaningful different features. Counter- and isofactuals can be used to identify positive and negative evidence in an image. This can also be visualized with heatmaps. We evaluate our approach against gradient-based attribution methods, which we find to produce meaningless adversarial perturbations. Using our method, we reveal biases in three different datasets. In a human subject experiment, we test whether non-experts find our method useful to spot spurious correlations learned by a model. Our work is a step towards more trustworthy explanations for computer vision. |
true | Improving Transformation Invariance in Contrastive Representation Learning
contrastive learning representation learning transformation invariance
We propose methods to strengthen the invariance properties of representations obtained by contrastive learning. While existing approaches implicitly induce a degree of invariance as representations are learned, we look to more directly enforce invariance in the encoding process. To this end, we first introduce a training objective for contrastive learning that uses a novel regularizer to control how the representation changes under transformation. We show that representations trained with this objective perform better on downstream tasks and are more robust to the introduction of nuisance transformations at test time. Second, we propose a change to how test time representations are generated by introducing a feature averaging approach that combines encodings from multiple transformations of the original input, finding that this leads to across the board performance gains. Finally, we introduce the novel Spirograph dataset to explore our ideas in the context of a differentiable generative process with multiple downstream tasks, showing that our techniques for learning invariance are highly beneficial. |
true | Reproducibility in Machine Learning for Health
reproducibility ML4H health systematic review replicability data access
Machine learning algorithms designed to characterize, monitor, and intervene on human health (ML4H) are expected to perform safely and reliably when operating at scale, potentially outside strict human supervision. This requirement warrants a stricter attention to issues of reproducibility than other fields of machine learning. In this work, we conduct a systematic evaluation of over 100 recently published ML4H research papers along several dimensions related to reproducibility we identified. We find that the field of ML4H compares poorly to more established machine learning fields, particularly concerning data accessibility and code accessibility. Finally, drawing from success in other fields of science, we propose recommendations to data providers, academic publishers, and the ML4H research community in order to promote reproducible research moving forward. |
false | Systematic Evaluation of Causal Discovery in Visual Model Based Reinforcement Learning
causal induction model based RL
Inducing causal relationships from observations is a classic problem in machine learning. Most work in causality starts from the premise that the causal variables themselves have known semantics or are observed. However, for AI agents such as robots trying to make sense of their environment, the only observables are low-level variables like pixels in images. To generalize well, an agent must induce high-level variables, particularly those which are causal or are affected by causal variables. A central goal for AI and causality is thus the joint discovery of abstract representations and causal structure. In this work, we systematically evaluate the agent's ability to learn underlying causal structure. We note that existing environments for studying causal induction are poorly suited for this objective because they have complicated task-specific causal graphs with many confounding factors. Hence, to facilitate research in learning the representation of high-level variables as well as causal structure among these variables, we present a suite of RL environments created to systematically probe the ability of methods to identify variables as well as causal structure among those variables. We evaluate various representation learning algorithms from literature and found that explicitly incorporating structure and modularity in the model can help causal induction in model-based reinforcement learning. |
false | Efficient Competitive Self-Play Policy Optimization
self-play policy optimization two-player zero-sum game multiagent
Reinforcement learning from self-play has recently reported many successes. Self-play, where the agents compete with themselves, is often used to generate training data for iterative policy improvement. In previous work, heuristic rules are designed to choose an opponent for the current learner. Typical rules include choosing the latest agent, the best agent, or a random historical agent. However, these rules may be inefficient in practice and sometimes do not guarantee convergence even in the simplest matrix games. This paper proposes a new algorithmic framework for competitive self-play reinforcement learning in two-player zero-sum games. We recognize the fact that the Nash equilibrium coincides with the saddle point of the stochastic payoff function, which motivates us to borrow ideas from classical saddle point optimization literature. Our method simultaneously trains several agents and intelligently takes each other as opponents based on a simple adversarial rule derived from a principled perturbation-based saddle optimization method. We prove theoretically that our algorithm converges to an approximate equilibrium with high probability in convex-concave games under standard assumptions. Beyond the theory, we further show the empirical superiority of our method over baseline methods relying on the aforementioned opponent-selection heuristics in matrix games, grid-world soccer, Gomoku, and simulated robot sumo, with neural net policy function approximators. |
true | Traversing Chemical Space with Latent Potential Flows
Molecule Optimization Generative Models Latent Traversal Optimal Transport
We consider the latent traversal problem in studying and exploring the chemical space with the learned latent space of a generative model. We propose a new framework, ChemFlow, that unified previous molecule manipulation and optimization method with a dynamical system perspective. Specifically, we formulate the problem as learning a vector field that transports the mass of the molecular distribution to the region with desired molecular properties or structure diversity. We also propose several alternative dynamics which exhibit various advantages over previous methods. We validate the efficacy of our proposed methods on both supervised and unsupervised molecule manipulation and optimization scenarios. |
true | Sub-Task Discovery with Limited Supervision: A Constrained Clustering Approach
discovery limited supervision constrained clustering hierarchical reinforcement captures information modular policies new tasks hierarchies policies
Hierarchical reinforcement learning captures sub-task information to learn modular policies that can be quickly adapted to new tasks. While hierarchies can be learned jointly with policies, this requires a lot of interaction. Traditional approaches require less data, but typically require sub-task labels to build a task hierarchy. We propose a semi-supervised constrained clustering approach to alleviate the labeling and interaction requirements. Our approach combines limited supervision with an arbitrary set of weak constraints, obtained purely from observations, that is jointly optimized to produce a clustering of the states into sub-tasks. We demonstrate improvement in two visual reinforcement learning tasks. |
false | Uniform Manifold Approximation with Two-phase Optimization
Dimensionality Reduction Manifold Learning Visualization Topological Data Analysis UMAP
We present a dimensionality reduction algorithm called Uniform Manifold Approximation with Two-phase Optimization (UMATO) which produces less biased global structures in the embedding results and is robust over diverse initialization methods than previous methods such as $t$-SNE and UMAP. We divide the optimization into two phases to alleviate the bias by establishing the global structure early using the representatives of the high-dimensional structures. The phases are 1) global optimization to obtain the overall skeleton of data and 2) local optimization to identify the regional characteristics of local areas. In our experiments with one synthetic and three real-world datasets, UMATO outperformed widely-used baseline algorithms, such as PCA, red, $t$-SNE, UMAP, topological autoencoders and Anchor $t$-SNE, in terms of quality metrics and 2D projection results. |
false | Efficient Generation of Diverse Scientific Hypotheses through Stepwise Conceptual Concretization
Hypothesis Generation Large Language Models AI for Science
In recent years, the automation of research using LLMs has been advancing rapidly. While The AI Scientist can generate papers that meet the acceptance criteria of top conferences in the machine learning field under specific conditions, there are limitations to the innovativeness of the generated research. As a step toward improving quality, this study aims to develop a method that generates scientific hypotheses of equivalent quality with significantly fewer tokens. The proposed method, which generates hypotheses more than ten times more efficiently, was compared with previous research in terms of novelty, importance, clarity, feasibility, and validity of the generated hypotheses. While no clear differences were observed in novelty and feasibility, improvements in performance were recognized in terms of importance, clarity, and validity compared to previous research. |
false | Design-Bench: Benchmarks for Data-Driven Offline Model-Based Optimization
Model-Based Optimization Benchmark Offline
Black-box model-based optimization (MBO) problems, where the goal is to find a design input that maximizes an unknown objective function, are ubiquitous in a wide range of domains, such as the design of drugs, aircraft, and robot morphology. Typically, such problems are solved by actively querying the black-box objective on design proposals and using the resulting feedback to improve the proposed designs. However, when the true objective function is expensive or dangerous to evaluate in the real world, we might instead prefer a method that can optimize this function using only previously collected data, for example from a set of previously conducted experiments. This data-driven offline MBO set- ting presents a number of unique challenges, but a number of recent works have demonstrated that viable offline MBO methods can be developed even for high- dimensional problems, using high-capacity deep neural network function approximators. Unfortunately, the lack of standardized evaluation tasks in this emerg- ing new field has made tracking progress and comparing recent methods difficult. To address this problem, we present Design-Bench, a benchmark suite of offline MBO tasks with a unified evaluation protocol and reference implementations of recent methods. Our benchmark suite includes diverse and realistic tasks derived from real-world problems in biology, material science, and robotics that present distinct challenges for offline MBO methods. Our benchmarks, together with the reference implementations, are available at sites.google.com/view/design-bench. We hope that our benchmark can serve as a meaningful metric for the progress of offline MBO methods and guide future algorithmic development. |
true | Local Search Algorithms for Rank-Constrained Convex Optimization
low rank rank-constrained convex optimization matrix completion
We propose greedy and local search algorithms for rank-constrained convex optimization, namely solving $\underset{\mathrm{rank}(A)\leq r^*}{\min}\, R(A)$ given a convex function $R:\mathbb{R}^{m\times n}\rightarrow \mathbb{R}$ and a parameter $r^*$. These algorithms consist of repeating two steps: (a) adding a new rank-1 matrix to $A$ and (b) enforcing the rank constraint on $A$. We refine and improve the theoretical analysis of Shalev-Shwartz et al. (2011), and show that if the rank-restricted condition number of $R$ is $\kappa$, a solution $A$ with rank $O(r^*\cdot \min\{\kappa \log \frac{R(\mathbf{0})-R(A^*)}{\epsilon}, \kappa^2\})$ and $R(A) \leq R(A^*) + \epsilon$ can be recovered, where $A^*$ is the optimal solution. This significantly generalizes associated results on sparse convex optimization, as well as rank-constrained convex optimization for smooth functions. We then introduce new practical variants of these algorithms that have superior runtime and recover better solutions in practice. We demonstrate the versatility of these methods on a wide range of applications involving matrix completion and robust principal component analysis.
|
true | Interactive Visual Exploration of Latent Space (IVELS) for peptide auto-encoder model selection
visualization peptide interactive auto-encoder vae
We present a tool for Interactive Visual Exploration of Latent Space (IVELS) for model selection. Evaluating generative models of discrete sequences from a continuous latent space is a challenging problem, since their optimization involves multiple competing objective terms. We introduce a model-selection pipeline to compare and filter models throughout consecutive stages of more complex and expensive metrics. We present the pipeline in an interactive visual tool to enable the exploration of the metrics, analysis of the learned latent space, and selection of the best model for a given task. We focus specifically on the variational auto-encoder family in a case study of modeling peptide sequences, which are short sequences of amino acids. This task is especially interesting due to the presence of multiple attributes we want to model. We demonstrate how an interactive visual comparison can assist in evaluating how well an unsupervised auto-encoder meaningfully captures the attributes of interest in its latent space. |
false | Diffusion Models in Space and Time via the Discretized Heat Equation
diffusion sde generative model pde
We propose a new class of diffusion models which use noising processes that diffuse jointly in space and time. These noising processes evolve according to a stochastic differential equation (SDE) inspired by the heat equation, a canonical space-time diffusion. We show that sampling from the diffusion’s transition density and evaluating its score remain tractable in the Fourier domain. This approach smooths the sequence of distributions that bridge noise and data, decaying high-frequency information before the lower frequencies that encode the large-scale structure of the image. We evaluate these models on MNIST and find that they generate convincing samples. |
true | Neural operators with localized integral and differential kernels
neural operators partial differential equations deep learning
Neural operators learn mappings between function spaces, which is applicable for learning solution operators of PDEs and other scientific modeling applications. Among them, the Fourier neural operator (FNO) is a popular architecture that performs global convolutions in the Fourier space. However, such global operations are often prone to over-smoothing and may fail to capture local details. In contrast, convolutional neural networks (CNN) can capture local features but are limited to training and inference at a single resolution. In this work, we present a principled approach to operator learning that can capture local features under two frameworks by learning differential operators and integral operators with locally supported kernels. Specifically, inspired by stencil methods, we prove that under an appropriate scaling of the kernel values of CNNs, we obtain differential operators. To obtain integral local operators, we utilize suitable basis representations for the kernels based on discrete-continuous convolutions. Both these principled approaches preserve the properties of operator learning and, hence, the ability to predict at any resolution. Adding our layers to FNOs significantly improves their performance, reducing the relative L2-error by 34-72% in our experiments on turbulent 2D Navier-Stokes fluid flow and the spherical shallow water equations. |
true | Solving Compositional Reinforcement Learning Problems via Task Reduction
compositional task sparse reward reinforcement learning task reduction imitation learning
We propose a novel learning paradigm, Self-Imitation via Reduction (SIR), for solving compositional reinforcement learning problems. SIR is based on two core ideas: task reduction and self-imitation. Task reduction tackles a hard-to-solve task by actively reducing it to an easier task whose solution is known by the RL agent. Once the original hard task is successfully solved by task reduction, the agent naturally obtains a self-generated solution trajectory to imitate. By continuously collecting and imitating such demonstrations, the agent is able to progressively expand the solved subspace in the entire task space. Experiment results show that SIR can significantly accelerate and improve learning on a variety of challenging sparse-reward continuous-control problems with compositional structures. Code and videos are available at https://sites.google.com/view/sir-compositional. |
false | A Transformer-based Framework for Multivariate Time Series Representation Learning
transformer multivariate time series unsupervised representation learning deep learning
In this work we propose for the first time a transformer-based framework for unsupervised representation learning of multivariate time series. Pre-trained models can be potentially used for downstream tasks such as regression and classification, forecasting and missing value imputation. We evaluate our models on several benchmark datasets for multivariate time series regression and classification and show that they exceed current state-of-the-art performance, even when the number of training samples is very limited, while at the same time offering computational efficiency. We show that unsupervised pre-training of our transformer models offers a substantial performance benefit over fully supervised learning, even without leveraging additional unlabeled data, i.e., by reusing the same data samples through the unsupervised objective. |
false | Similarity Search for Efficient Active Learning and Search of Rare Concepts
active learning active search
Many active learning and search approaches are intractable for industrial settings with billions of unlabeled examples. Existing approaches, such as uncertainty sampling or information density, search globally for the optimal examples to label, scaling linearly or even quadratically with the unlabeled data. However, in practice, data is often heavily skewed; only a small fraction of collected data will be relevant for a given learning task. For example, when identifying rare classes, detecting malicious content, or debugging model performance, positive examples can appear in less than 1% of the data. In this work, we exploit this skew in large training datasets to reduce the number of unlabeled examples considered in each selection round by only looking at the nearest neighbors to the labeled examples. Empirically, we observe that learned representations can effectively cluster unseen concepts, making active learning very effective and substantially reducing the number of viable unlabeled examples. We evaluate several selection strategies in this setting on three large-scale computer vision datasets: ImageNet, OpenImages, and a proprietary dataset of 10 billion images from a large internet company. For rare classes, active learning methods need as little as 0.31% of the labeled data to match the average precision of full supervision. By limiting the selection strategies to the immediate neighbors of the labeled data as candidates for labeling, we process as little as 0.1% of the unlabeled data while achieving similar reductions in labeling costs as the traditional global approach. This process of expanding the candidate pool with the nearest neighbors of the labeled set can be done efficiently and reduces the computational complexity of selection by orders of magnitude. |
false | The large learning rate phase of deep learning
deep learning wide networks training dynamics
The choice of initial learning rate can have a profound effect on the performance of deep networks. We present empirical evidence that networks exhibit sharply distinct behaviors at small and large learning rates. In the small learning rate phase, training can be understood using the existing theory of infinitely wide neural networks. At large learning rates, we find that networks exhibit qualitatively distinct phenomena that cannot be explained by existing theory: The loss grows during the early part of training, and optimization eventually converges to a flatter minimum. Furthermore, we find that the optimal performance is often found in the large learning rate phase. To better understand this behavior we analyze the dynamics of a two-layer linear network and prove that it exhibits these different phases. We find good agreement between our analysis and the training dynamics observed in realistic deep learning settings.
|
true | Practical Real Time Recurrent Learning with a Sparse Approximation
recurrent neural networks backpropagation biologically plausible forward mode real time recurrent learning rtrl bptt
Recurrent neural networks are usually trained with backpropagation through time, which requires storing a complete history of network states, and prohibits updating the weights "online" (after every timestep). Real Time Recurrent Learning (RTRL) eliminates the need for history storage and allows for online weight updates, but does so at the expense of computational costs that are quartic in the state size. This renders RTRL training intractable for all but the smallest networks, even ones that are made highly sparse.
We introduce the Sparse n-step Approximation (SnAp) to the RTRL influence matrix. SnAp only tracks the influence of a parameter on hidden units that are reached by the computation graph within $n$ timesteps of the recurrent core. SnAp with $n=1$ is no more expensive than backpropagation but allows training on arbitrarily long sequences. We find that it substantially outperforms other RTRL approximations with comparable costs such as Unbiased Online Recurrent Optimization. For highly sparse networks, SnAp with $n=2$ remains tractable and can outperform backpropagation through time in terms of learning speed when updates are done online. |
false | Federated Learning for Decentralized Scientific Collaboration: Privacy-Preserving Multi-Agent AI for Cross-Domain Research
Federated Learning Multi-Agent Systems Privacy-Preserving AI Decentralized Collaboration Secure Aggregation Differential Privacy Meta-Learning Scientific AI Cross-Domain Knowledge Transfer Hierarchical FL
Scientific collaboration often requires multi-institutional AI training, yet privacy concerns, regulatory constraints, and data heterogeneity hinder centralized model development. This paper introduces a federated learning (FL) framework that enables scientific agents to collaboratively refine AI models without sharing raw data. By integrating secure aggregation, differential privacy, and multi-agent orchestration, the system ensures efficient cross-domain knowledge transfer in applications like genomics, medical research, and climate science.
Proposed method achieves 35% faster model convergence compared to single-institution baselines, validated with $p < 0.05$, while maintaining low privacy leakage risk. Unlike traditional FL, our framework incorporates agentic AI coordination, allowing domain-specific adaptation and conflict resolution across institutions. We discuss scalability challenges, propose hierarchical FL solutions, and outline future work in theoretical guarantees and real-world deployment.
This approach presents a scalable and privacy-preserving alternative to centralized AI training, accelerating scientific discovery while respecting data sovereignty. |
true | AUXILIARY TASK UPDATE DECOMPOSITION: THE GOOD, THE BAD AND THE NEUTRAL
pre-training multitask learning deeplearning gradient decomposition
While deep learning has been very beneficial in data-rich settings, tasks with smaller training set
often resort to pre-training or multitask learning to leverage data from other tasks. In this case,
careful consideration is needed to select tasks and model parameterizations such that updates from
the auxiliary tasks actually help the primary task. We seek to alleviate this burden by formulating a model-agnostic framework that performs fine-grained manipulation of the auxiliary task gradients. We propose to decompose auxiliary updates into directions which help, damage or leave the primary task loss unchanged. This allows weighting the update directions
differently depending on their impact on the problem of interest. We present a novel and efficient algorithm for that
purpose and show its advantage in practice. Our method leverages efficient automatic differentiation
procedures and randomized singular value decomposition for scalability. We show that our framework is
generic and encompasses some prior work as particular cases. Our approach consistently outperforms strong and widely used baselines when leveraging out-of-distribution data for Text and Image classification tasks. |
true | Robust Reinforcement Learning on State Observations with Learned Optimal Adversary
reinforcement learning robustness adversarial attacks adversarial defense
We study the robustness of reinforcement learning (RL) with adversarially perturbed state observations, which aligns with the setting of many adversarial attacks to deep reinforcement learning (DRL) and is also important for rolling out real-world RL agent under unpredictable sensing noise. With a fixed agent policy, we demonstrate that an optimal adversary to perturb state observations can be found, which is guaranteed to obtain the worst case agent reward. For DRL settings, this leads to a novel empirical adversarial attack to RL agents via a learned adversary that is much stronger than previous ones. To enhance the robustness of an agent, we propose a framework of alternating training with learned adversaries (ATLA), which trains an adversary online together with the agent using policy gradient following the optimal adversarial attack framework. Additionally, inspired by the analysis of state-adversarial Markov decision process (SA-MDP), we show that past states and actions (history) can be useful for learning a robust agent, and we empirically find a LSTM based policy can be more robust under adversaries. Empirical evaluations on a few continuous control environments show that ATLA achieves state-of-the-art performance under strong adversaries. Our code is available at https://github.com/huanzhang12/ATLA_robust_RL. |
false | Empirical Frequentist Coverage of Deep Learning Uncertainty Quantification Procedures
uncertainty quantification coverage dataset shift
Uncertainty quantification for complex deep learning models is increasingly important as these techniques see growing use in high-stakes, real-world settings. Currently, the quality of a model's uncertainty is evaluated using point-prediction metrics such as negative log-likelihood or the Brier score on heldout data. In this study, we provide the first large scale evaluation of the empirical frequentist coverage properties of well known uncertainty quantification techniques on a suite of regression and classification tasks. We find that, in general, some methods do achieve desirable coverage properties on \emph{in distribution} samples, but that coverage is not maintained on out-of-distribution data. Our results demonstrate the failings of current uncertainty quantification techniques as dataset shift increases and establish coverage as an important metric in developing models for real-world applications. |
true | An Understanding of Learning from Demonstrations for Neural Text Generation
Text Generation Reinforcement Learning Learning from Demonstrations
In this blog post, we will go over the ICLR 2021 paper titled Text Generation by Learning from Demonstrations. This paper introduces a learning method based on offline, off-policy reinforcement learning (RL) which addresses two key limitations of a training objective used in neural text generation models: Maximum Likelihood Estimate (MLE).
Goal of this blog post: Our main goal with this blog post is to provide researchers and practitioners in both NLP and RL with (1) a better understanding of algorithm presented in this paper (GOLD), and (2) an understanding of how RL is used for text generation. |
true | Certify or Predict: Boosting Certified Robustness with Compositional Architectures
Provable Robustness Network Architecture Robustness Adversarial Accuracy Certified Robustness
A core challenge with existing certified defense mechanisms is that while they improve certified robustness, they also tend to drastically decrease natural accuracy, making it difficult to use these methods in practice. In this work, we propose a new architecture which addresses this challenge and enables one to boost the certified robustness of any state-of-the-art deep network, while controlling the overall accuracy loss, without requiring retraining. The key idea is to combine this model with a (smaller) certified network where at inference time, an adaptive selection mechanism decides on the network to process the input sample. The approach is compositional: one can combine any pair of state-of-the-art (e.g., EfficientNet or ResNet) and certified networks, without restriction. The resulting architecture enables much higher natural accuracy than previously possible with certified defenses alone, while substantially boosting the certified robustness of deep networks. We demonstrate the effectiveness of this adaptive approach on a variety of datasets and architectures. For instance, on CIFAR-10 with an $\ell_\infty$ perturbation of 2/255, we are the first to obtain a high natural accuracy (90.1%) with non-trivial certified robustness (27.5%). Notably, prior state-of-the-art methods incur a substantial drop in accuracy for a similar certified robustness. |
true | Semi-supervised Keypoint Localization
semi-supervised learning keypoint localization limited data unsupervised loss
Knowledge about the locations of keypoints of an object in an image can assist in fine-grained classification and identification tasks, particularly for the case of objects that exhibit large variations in poses that greatly influence their visual appearance, such as wild animals. However, supervised training of a keypoint detection network requires annotating a large image dataset for each animal species, which is a labor-intensive task. To reduce the need for labeled data, we propose to learn simultaneously keypoint heatmaps and pose invariant keypoint representations in a semi-supervised manner using a small set of labeled images along with a larger set of unlabeled images. Keypoint representations are learnt with a semantic keypoint consistency constraint that forces the keypoint detection network to learn similar features for the same keypoint across the dataset. Pose invariance is achieved by making keypoint representations for the image and its augmented copies closer together in feature space. Our semi-supervised approach significantly outperforms previous methods on several benchmarks for human and animal body landmark localization. |
true | ENHANCING DIVERSITY AND NOVELTY IN TEXT GENERATION VIA MULTI-VIEW EMBEDDINGS
Diversity and Novelty Large Language Models Multiview Embeddings
Large Language Models (LLMs) demonstrate remarkable proficiency in generating accurate and fluent text. However, they often struggle with diversity and novelty, leading to repetitive or overly deterministic responses. These limitations stem from constraints in training data, including gaps in specific knowledge domains, outdated information, and an over-reliance on textual sources. Such shortcomings reduce their effectiveness in tasks requiring creativity, multi-perspective reasoning, and exploratory thinking. To address this challenge, we introduce multi-view embeddings, a novel approach that enriches input prompts with diverse perspectives derived from both textual and visual sources. By incorporating additional contextual information, this method enhances the variety and creativity of generated outputs. Importantly, our approach is model-agnostic, requiring no architectural modifications and being compatible with both open-source and proprietary LLMs.
Furthermore, we propose a comprehensive evaluation framework that simultaneously measures diversity, novelty, and correctness—a first-of-its-kind methodology for assessing these three crucial aspects of LLM-generated content. We evaluate our method and framework on over 469,000 generated outputs from various well-known LLMs, demonstrating significant improvements in output diversity and novelty while maintaining quality and relevance. Our approach provides a scalable and practical solution for enhancing LLM performance across a wide range of applications, including brainstorming, creative writing, and multiple-choice question generation. |
false | ChemistryQA: A Complex Question Answering Dataset from Chemistry
kind qa tasks chemistryqa complex question dataset neural networks chemistry chemistryqa chemistry many question tasks
Many Question Answering (QA) tasks have been studied in NLP and employed to evaluate the progress of machine intelligence. One kind of QA tasks, such as Machine Reading Comprehension QA, is well solved by end-to-end neural networks; another kind of QA tasks, such as Knowledge Base QA, needs to be translated to a formatted representations and then solved by a well-designed solver. We notice that some real-world QA tasks are more complex, which cannot be solved by end-to-end neural networks or translated to any kind of formal representations. To further stimulate the research of QA and development of QA techniques, in this work, we create a new and complex QA dataset, ChemistryQA, based on real-world chemical calculation questions. To answer chemical questions, machines need to understand questions, apply chemistry and Math knowledge, and do calculation and reasoning. To help researchers ramp up, we build two baselines: the first one is BERT-based sequence to sequence model, and the second one is an extraction system plus a graph search based solver. These two methods achieved 0.164 and 0.169 accuracy on the development set, respectively, which clearly demonstrate that new techniques are needed for complex QA tasks. ChemistryQA dataset will be available for public download once the paper is published. |
false | Representation and Bias in Multilingual NLP: Insights from Controlled Experiments on Conditional Language Modeling
multilinguality science for NLP fundamental science in the era of AI/DL representation learning for language conditional language modeling Transformer Double Descent non-monotonicity fairness meta evaluation visualization or interpretation of learned representations
Inspired by the phenomenon of performance disparity between languages in machine translation, we investigate whether and to what extent languages are equally hard to "conditional-language-model". Our goal is to improve our understanding and expectation of the relationship between language, data representation, size, and performance. We study one-to-one, bilingual conditional language modeling through a series of systematically controlled experiments with the Transformer and the 6 languages from the United Nations Parallel Corpus. We examine character, byte, and word models in 30 language directions and 5 data sizes, and observe indications suggesting a script bias on the character level, a length bias on the byte level, and a word bias that gives rise to a hierarchy in performance across languages. We also identify two types of sample-wise non-monotonicity --- while word-based representations are prone to exhibit Double Descent, length can induce unstable performance across the size range studied in a novel meta phenomenon which we term "erraticity". By eliminating statistically significant performance disparity on the character and byte levels by normalizing length and vocabulary in the data, we show that, in the context of computing with the Transformer, there is no complexity intrinsic to languages other than that related to their statistical attributes and that performance disparity is not a necessary condition but a byproduct of word segmentation. Our application of statistical comparisons as a fairness measure also serves as a novel rigorous method for the intrinsic evaluation of languages, resolving a decades-long debate on language complexity. While these quantitative biases leading to disparity are mitigable through a shallower network, we find room for a human bias to be reflected upon. We hope our work helps open up new directions in the area of language and computing that would be fairer and more flexible and foster a new transdisciplinary perspective for DL-inspired scientific progress. |
true | Structured Prediction as Translation between Augmented Natural Languages
language models few-shot learning transfer learning structured prediction generative modeling sequence to sequence multi-task learning
We propose a new framework, Translation between Augmented Natural Languages (TANL), to solve many structured prediction language tasks including joint entity and relation extraction, nested named entity recognition, relation classification, semantic role labeling, event extraction, coreference resolution, and dialogue state tracking. Instead of tackling the problem by training task-specific discriminative classifiers, we frame it as a translation task between augmented natural languages, from which the task-relevant information can be easily extracted. Our approach can match or outperform task-specific models on all tasks, and in particular achieves new state-of-the-art results on joint entity and relation extraction (CoNLL04, ADE, NYT, and ACE2005 datasets), relation classification (FewRel and TACRED), and semantic role labeling (CoNLL-2005 and CoNLL-2012). We accomplish this while using the same architecture and hyperparameters for all tasks, and even when training a single model to solve all tasks at the same time (multi-task learning). Finally, we show that our framework can also significantly improve the performance in a low-resource regime, thanks to better use of label semantics. |
End of preview. Expand
in Data Studio
README.md exists but content is empty.
- Downloads last month
- 102