id
float64 2.21k
2.31k
| content
stringlengths 78
1.92k
⌀ | category
stringclasses 8
values |
|---|---|---|
2,312.06515 |
The threat of hardware Trojans (HTs) in security-critical IPs like cryptographic accelerators poses severe security risks. The HT detection methods available today mostly rely on golden models and detailed circuit specifications. Often they are specific to certain HT payload types, making pre-silicon verification difficult and leading to security gaps. We propose a novel formal verification method for HT detection in non-interfering accelerators at the Register Transfer Level (RTL), employing standard formal property checking. Our method guarantees the exhaustive detection of any sequential HT independently of its payload behavior, including physical side channels. It does not require a golden model or a functional specification of the design. The experimental results demonstrate efficient and effective detection of all sequential HTs in accelerators available on Trust-Hub, including those with complex triggers and payloads.
|
cs
|
2,312.03309 |
Continual learning addresses the problem of continuously acquiring and transferring knowledge without catastrophic forgetting of old concepts. While humans achieve continual learning via diverse neurocognitive mechanisms, there is a mismatch between cognitive properties and evaluation methods of continual learning models. First, the measurement of continual learning models mostly relies on evaluation metrics at a micro-level, which cannot characterize cognitive capacities of the model. Second, the measurement is method-specific, emphasizing model strengths in one aspect while obscuring potential weaknesses in other respects. To address these issues, we propose to integrate model cognitive capacities and evaluation metrics into a unified evaluation paradigm. We first characterize model capacities via desiderata derived from cognitive properties supporting human continual learning. The desiderata concern (1) adaptability in varying lengths of task sequence; (2) sensitivity to dynamic task variations; and (3) efficiency in memory usage and training time consumption. Then we design evaluation protocols for each desideratum to assess cognitive capacities of recent continual learning models. Experimental results show that no method we consider has satisfied all the desiderata and is still far away from realizing truly continual learning. Although some methods exhibit some degree of adaptability and efficiency, no method is able to identify task relationships when encountering dynamic task variations, or achieve a trade-off in learning similarities and differences between tasks. Inspired by these results, we discuss possible factors that influence model performance in these desiderata and provide guidance for the improvement of continual learning models.
|
cs
|
2,312.04292 |
An explanation is proposed for the appearance of slowly drift shadow bursts in the dynamic spectrum of Jupiter against the background of decameter radio emission with a quasi-harmonic structure. Background radio emission is caused by hot ions with a loss cone type distribution function, which generate ion cyclotron waves due to the effect of double plasma resonance. A flow of hot ions with a distribution function of the Maxwell type is injected into the source region, fills the loss cone of generating ions and interrupts the generation of ion cyclotron waves due to the filling of the loss cone. The condition under which instability breaks down is obtained, and the optimal values of the parameters of the injected ions necessary for the occurrence of bursts in absorption are determined.
|
astro-ph
|
2,312.03457 |
We compute the class groups of full rank upper cluster algebras in terms of the exchange polynomials. This characterizes the UFDs among these algebras. Our results simultaneously generalize theorems of Garcia Elsener, Lampe, and Smertnig from 2019 and of Cao, Keller, and Qin from 2023. Furthermore we show that every (upper) cluster algebra is a finite factorization domain.
|
math
|
2,312.06104 |
Constraint satisfaction problems are an important area of computer science. Many of these problems are in the complexity class NP which is exponentially hard for all known methods, both for worst cases and often typical. Fundamentally, the lack of any guided local minimum escape method ensures the hardness of both exact and approximate optimization classically, but the intuitive mechanism for approximation hardness in quantum algorithms based on Hamiltonian time evolution is poorly understood. We explore this question using the prototypically hard MAX-3-XORSAT problem class. We conclude that the mechanisms for quantum exact and approximation hardness are fundamentally distinct. We qualitatively identify why traditional methods such as quantum adiabatic optimization are not good approximation algorithms. We propose a new spectral folding optimization method that does not suffer from these issues and study it analytically and numerically. We consider random rank-3 hypergraphs including extremal planted solution instances, where the ground state satisfies an anomalously high fraction of constraints compared to truly random problems. We show that, if we define the energy to be $E = N_{unsat}-N_{sat}$, then spectrally folded quantum optimization will return states with energy $E \leq A E_{GS}$ (where $E_{GS}$ is the ground state energy) in polynomial time, where conservatively, $A \simeq 0.6$. We thoroughly benchmark variations of spectrally folded quantum optimization for random classically approximation-hard (planted solution) instances in simulation, and find performance consistent with this prediction. We do not claim that this approximation guarantee holds for all possible hypergraphs, though our algorithm's mechanism can likely generalize widely. These results suggest that quantum computers are more powerful for approximate optimization than had been previously assumed.
|
quant-ph
|
2,312.03145 |
Associated to each graph G is a Gaussian graphical model. Such models are often used in high-dimensional settings, i.e. where there are relatively few data points compared to the number of variables. The maximum likelihood threshold of a graph is the minimum number of data points required to fit the corresponding graphical model using maximum likelihood estimation. Graphical lasso is a method for selecting and fitting a graphical model. In this project, we ask: when graphical lasso is used to select and fit a graphical model on n data points, how likely is it that n is greater than or equal to the maximum likelihood threshold of the corresponding graph? Our results are a series of computational experiments.
|
math
|
2,312.04952 |
We introduce a natural notion of mean (or average) distance in the context of compact metric graphs, and study its relation to geometric properties of the graph. We show that it exhibits a striking number of parallels to the reciprocal of the spectral gap of the graph Laplacian with standard vertex conditions: it is maximised among all graphs of fixed length by the path graph (interval), or by the loop in the restricted class of doubly connected graphs, and it is minimised among all graphs of fixed length and number of edges by the equilateral flower graph. We also establish bounds for the correctly scaled product of the spectral gap and the square of the mean distance which depend only on combinatorial, and not metric, features of the graph. This raises the open question whether this product admits absolute upper and lower bounds valid on all compact metric graphs.
|
math
|
2,312.0309 |
Computing artifacts tend to exclude marginalized students, so we must create new methods to critique and change them. We studied the potential for "satirical programming" to critique artifacts as part of culturally responsive computing (CRC) pedagogy. We conducted a one-hour session for three different BPC programs (N=51). We showed an example of a satirical Python script and taught elements of Python to create a script. Our findings suggest this method is a promising CRC pedagogical approach: 50% of marginalized students worked together to create a satirical script, and 80% enjoyed translating their "glitches" into satirical Python scripts.
|
cs
|
2,312.02749 |
In this paper, we provide formulas calculating the partition functions of two types of plane partitions using the crystal melting model method introduced by Okounkov, Reshetikhin and Vafa. As applications, we obtain a product formula for the partition function of the plane partitions with a limit shape boundary. A corollary of this formula is the demonstration of the equivalence between this partition function and the open-closed string amplitude of the double$-\mathbb{P}^1$ model. We also derive a product formula for the partition function of symmetric plane partitions with a limit shape boundary.
|
math
|
2,312.03668 |
Advances in machine learning have made it possible to perform various text and speech processing tasks, including automatic speech recognition (ASR), in an end-to-end (E2E) manner. Since typical E2E approaches require large amounts of training data and resources, leveraging pre-trained foundation models instead of training from scratch is gaining attention. Although there have been attempts to use pre-trained speech and language models in ASR, most of them are limited to using either. This paper explores the potential of integrating a pre-trained speech representation model with a large language model (LLM) for E2E ASR. The proposed model enables E2E ASR by generating text tokens in an autoregressive manner via speech representations as speech prompts, taking advantage of the vast knowledge provided by the LLM. Furthermore, the proposed model can incorporate remarkable developments for LLM utilization, such as inference optimization and parameter-efficient domain adaptation. Experimental results show that the proposed model achieves performance comparable to modern E2E ASR models.
|
cs
|
2,312.04603 |
Online polarization has attracted the attention of researchers for many years. Its effects on society are a cause for concern, and the design of personalized depolarization strategies appears to be a key solution. Such strategies should rely on a fine and accurate measurement, and a clear understanding of polarization behaviors. However, the literature still lacks ways to characterize them finely. We propose GRAIL, the first individual polarization metric, relying on multiple factors. GRAIL assesses these factors through entropy and is based on an adaptable Generalized Additive Model. We evaluate the proposed metric on a Twitter dataset related to the highly controversial debate about the COVID-19 vaccine. Experiments confirm the ability of GRAIL to discriminate between polarization behaviors. To go further, we provide a finer characterization and explanation of the identified behaviors through an innovative evaluation framework.
|
cs
|
2,312.03187 |
Researchers have proposed to use data of human preference feedback to fine-tune text-to-image generative models. However, the scalability of human feedback collection has been limited by its reliance on manual annotation. Therefore, we develop and test a method to automatically annotate user preferences from their spontaneous facial expression reaction to the generated images. We collect a dataset of Facial Expression Reaction to Generated Images (FERGI) and show that the activations of multiple facial action units (AUs) are highly correlated with user evaluations of the generated images. Specifically, AU4 (brow lowerer) is most consistently reflective of negative evaluations of the generated image. This can be useful in two ways. Firstly, we can automatically annotate user preferences between image pairs with substantial difference in AU4 responses to them with an accuracy significantly outperforming state-of-the-art scoring models. Secondly, directly integrating the AU4 responses with the scoring models improves their consistency with human preferences. Additionally, the AU4 response best reflects the user's evaluation of the image fidelity, making it complementary to the state-of-the-art scoring models, which are generally better at reflecting image-text alignment. Finally, this method of automatic annotation with facial expression analysis can be potentially generalized to other generation tasks. The code is available at , and the dataset is also available at the same link for research purposes.
|
cs
|
2,312.04734 |
Recurrence is a fundamental characteristic of complicated deterministic dynamical systems. Understanding its structure is challenging, especially if the system of interest has many degrees of freedom so that visualizations of trajectories are of limited use. To analyze recurrent phenomena, we propose a computational method to identify oscillations and the transitions between them. To this end, we introduce the concept of cycling signature, which is a topological descriptor of a trajectory segment. The advantage of a topological approach, in particular in an application setting, is that it is robust to noise. Cycling signatures are computable from data and provide a comprehensive global description of oscillations through their statistics over many trajectory segments. We demonstrate this through three examples. In particular, we identify and analyze six oscillations in a four-dimensional system with a hyperchaotic attractor.
|
math
|
2,312.05338 |
Robotic-based compact storage and retrieval systems provide high-density storage in distribution center and warehouse applications. In the system, items are stored in bins, and the bins are organized inside a three-dimensional grid. Robots move on top of the grid to retrieve and deliver bins. To retrieve a bin, a robot removes all bins above one by one with its gripper, called bin digging. The closer the target bin is to the top of the grid, the less digging is required to retrieve the bin. In this paper, we propose a policy to optimally arrange the bins in the grid while processing bin requests so that the most frequently accessed bins remain near the top of the grid. This improves the performance of the system and makes it responsive to changes in bin demand. Our solution approach identifies the optimal bin arrangement in the storage facility, initiates a transition to this optimal set-up, and subsequently ensures the ongoing maintenance of this arrangement for optimal performance. We perform extensive simulations on a custom-built discrete event model of the system. Our simulation results show that under the proposed policy more than half of the bins requested are located on top of the grid, reducing bin digging compared to existing policies. Compared to existing approaches, the proposed policy reduces the retrieval time of the requested bins by over 30% and the number of bin requests that exceed certain time thresholds by nearly 50%.
|
cs
|
2,312.02167 |
Recent studies have confirmed cardiovascular diseases remain responsible for highest death toll amongst non-communicable diseases. Accurate left ventricular (LV) volume estimation is critical for valid diagnosis and management of various cardiovascular conditions, but poses significant challenge due to inherent uncertainties associated with segmentation algorithms in magnetic resonance imaging (MRI). Recent machine learning advancements, particularly U-Net-like convolutional networks, have facilitated automated segmentation for medical images, but struggles under certain pathologies and/or different scanner vendors and imaging protocols. This study proposes a novel methodology for post-hoc uncertainty estimation in LV volume prediction using Itô stochastic differential equations (SDEs) to model path-wise behavior for the prediction error. The model describes the area of the left ventricle along the heart's long axis. The method is agnostic to the underlying segmentation algorithm, facilitating its use with various existing and future segmentation technologies. The proposed approach provides a mechanism for quantifying uncertainty, enabling medical professionals to intervene for unreliable predictions. This is of utmost importance in critical applications such as medical diagnosis, where prediction accuracy and reliability can directly impact patient outcomes. The method is also robust to dataset changes, enabling application for medical centers with limited access to labeled data. Our findings highlight the proposed uncertainty estimation methodology's potential to enhance automated segmentation robustness and generalizability, paving the way for more reliable and accurate LV volume estimation in clinical settings as well as opening new avenues for uncertainty quantification in biomedical image segmentation, providing promising directions for future research.
|
cs
|
2,312.05647 |
Exact methods for exponentiation of matrices of dimension $N$ can be computationally expensive in terms of execution time ($N^{3}$) and memory requirements ($N^{2}$) not to mention numerical precision issues. A type of matrix often exponentiated in the sciences is the rate matrix. Here we explore five methods to exponentiate rate matrices some of which apply even more broadly to other matrix types. Three of the methods leverage a mathematical analogy between computing matrix elements of a matrix exponential and computing transition probabilities of a dynamical processes (technically a Markov jump process, MJP, typically simulated using Gillespie). In doing so, we identify a novel MJP-based method relying on restricting the number of ``trajectory" jumps based on the magnitude of the matrix elements with favorable computational scaling. We then discuss this method's downstream implications on mixing properties of Monte Carlo posterior samplers. We also benchmark two other methods of matrix exponentiation valid for any matrix (beyond rate matrices and, more generally, positive definite matrices) related to solving differential equations: Runge-Kutta integrators and Krylov subspace methods. Under conditions where both the largest matrix element and the number of non-vanishing elements scale linearly with $N$ -- reasonable conditions for rate matrices often exponentiated -- computational time scaling with the most competitive methods (Krylov and one of the MJP-based methods) reduces to $N^2$ with total memory requirements of $N$.
|
cs
|
2,312.03329 |
We consider an intermediate factor situation in two categories: probability measure preserving ergodic theory and compact topological dynamics. In the first we prove a master-key theorem and examine a wide range of applications. In the second we provide some counterexamples.
|
math
|
2,312.0231 |
Recent advancements in language-model-based video understanding have been progressing at a remarkable pace, spurred by the introduction of Large Language Models (LLMs). However, the focus of prior research has been predominantly on devising a projection layer that maps video features to tokens, an approach that is both rudimentary and inefficient. In our study, we introduce a cutting-edge framework, VaQuitA, designed to refine the synergy between video and textual information. At the data level, instead of sampling frames uniformly, we implement a sampling method guided by CLIP-score rankings, which enables a more aligned selection of frames with the given question. At the feature level, we integrate a trainable Video Perceiver alongside a Visual-Query Transformer (abbreviated as VQ-Former), which bolsters the interplay between the input question and the video features. We also discover that incorporating a simple prompt, "Please be critical", into the LLM input can substantially enhance its video comprehension capabilities. Our experimental results indicate that VaQuitA consistently sets a new benchmark for zero-shot video question-answering tasks and is adept at producing high-quality, multi-turn video dialogues with users.
|
cs
|
2,312.04563 |
Structure-from-motion (SfM) is a long-standing problem in the computer vision community, which aims to reconstruct the camera poses and 3D structure of a scene from a set of unconstrained 2D images. Classical frameworks solve this problem in an incremental manner by detecting and matching keypoints, registering images, triangulating 3D points, and conducting bundle adjustment. Recent research efforts have predominantly revolved around harnessing the power of deep learning techniques to enhance specific elements (e.g., keypoint matching), but are still based on the original, non-differentiable pipeline. Instead, we propose a new deep pipeline VGGSfM, where each component is fully differentiable and thus can be trained in an end-to-end manner. To this end, we introduce new mechanisms and simplifications. First, we build on recent advances in deep 2D point tracking to extract reliable pixel-accurate tracks, which eliminates the need for chaining pairwise matches. Furthermore, we recover all cameras simultaneously based on the image and track features instead of gradually registering cameras. Finally, we optimise the cameras and triangulate 3D points via a differentiable bundle adjustment layer. We attain state-of-the-art performance on three popular datasets, CO3D, IMC Phototourism, and ETH3D.
|
cs
|
2,312.03518 |
In this paper, we enhance a recent algorithm for approximate spectral factorization of matrix functions, extending its capabilities to precisely factorize rational matrices when an exact lower-upper triangular factorization is available. This novel approach leverages a fundamental component of the improved algorithm for the precise design of rational paraunitary filter banks, allowing for the predetermined placement of zeros and poles. The introduced algorithm not only advances the state-of-the-art in spectral factorization but also opens new avenues for the tailored design of paraunitary filters with specific spectral properties, offering significant potential for applications in signal processing and beyond.
|
math
|
2,312.03762 |
We explore colour versus shape goal misgeneralization originally demonstrated by Di Langosco et al. (2022) in the Procgen Maze environment, where, given an ambiguous choice, the agents seem to prefer generalization based on colour rather than shape. After training over 1,000 agents in a simplified version of the environment and evaluating them on over 10 million episodes, we conclude that the behaviour can be attributed to the agents learning to detect the goal object through a specific colour channel. This choice is arbitrary. Additionally, we show how, due to underspecification, the preferences can change when retraining the agents using exactly the same procedure except for using a different random seed for the training run. Finally, we demonstrate the existence of outliers in out-of-distribution behaviour based on training random seed alone.
|
cs
|
2,312.04217 |
We present a hybrid method for time-dependent particle transport problems that combines Monte Carlo (MC) estimation with deterministic solutions based on discrete ordinates. For spatial discretizations, the MC algorithm computes a piecewise constant solution and the discrete ordinates uses bilinear discontinuous finite elements. From the hybridization of the problem, the resulting problem solved by Monte Carlo is scattering free, resulting in a simple, efficient solution procedure. Between time steps, we use a projection approach to ``relabel'' collided particles as uncollided particles. From a series of standard 2-D Cartesian test problems we observe that our hybrid method has improved accuracy and reduction in computational complexity of approximately an order of magnitude relative to standard discrete ordinates solutions.
|
cs
|
2,312.05734 |
Learning methods in Banach spaces are often formulated as regularization problems which minimize the sum of a data fidelity term in a Banach norm and a regularization term in another Banach norm. Due to the infinite dimensional nature of the space, solving such regularization problems is challenging. We construct a direct sum space based on the Banach spaces for the data fidelity term and the regularization term, and then recast the objective function as the norm of a suitable quotient space of the direct sum space. In this way, we express the original regularized problem as an unregularized problem on the direct sum space, which is in turn reformulated as a dual optimization problem in the dual space of the direct sum space. The dual problem is to find the maximum of a linear function on a convex polytope, which may be solved by linear programming. A solution of the original problem is then obtained by using related extremal properties of norming functionals from a solution of the dual problem. Numerical experiments are included to demonstrate that the proposed duality approach leads to an implementable numerical method for solving the regularization learning problems.
|
math
|
2,312.06135 |
Artistic style transfer aims to repaint the content image with the learned artistic style. Existing artistic style transfer methods can be divided into two categories: small model-based approaches and pre-trained large-scale model-based approaches. Small model-based approaches can preserve the content strucuture, but fail to produce highly realistic stylized images and introduce artifacts and disharmonious patterns; Pre-trained large-scale model-based approaches can generate highly realistic stylized images but struggle with preserving the content structure. To address the above issues, we propose ArtBank, a novel artistic style transfer framework, to generate highly realistic stylized images while preserving the content structure of the content images. Specifically, to sufficiently dig out the knowledge embedded in pre-trained large-scale models, an Implicit Style Prompt Bank (ISPB), a set of trainable parameter matrices, is designed to learn and store knowledge from the collection of artworks and behave as a visual prompt to guide pre-trained large-scale models to generate highly realistic stylized images while preserving content structure. Besides, to accelerate training the above ISPB, we propose a novel Spatial-Statistical-based self-Attention Module (SSAM). The qualitative and quantitative experiments demonstrate the superiority of our proposed method over state-of-the-art artistic style transfer methods.
|
cs
|
2,312.05401 |
In this work, we present a method for turning Chinese still-life paintings with global illumination effects into dynamic paintings with moving lights. Our goal is to preserve the original look and feel of still-life paintings with moving lights and objects. We have developed a deceptively simple method that can be computed as a composite of two animated texture images using an animated rendering. The compositing process can be implemented directly in an animation system such as AfterEffect, which allows for the basic compositing operation over animations. It is also possible to control the colors by changing the material colors in animated rendering. We have provided a proof-of-concept based on an original digital Still-Life painting that is in realist Chinese style. This approach can be used to turn almost any still-life painting into a dynamic painting.
|
cs
|
2,312.03148 |
Theta Mus is a remarkable spectroscopic binary (SB) consisting of a carbon-type Wolf-Rayet star and OV companion (WC6+O6-7V) in a 19-day orbit. In addition an O-supergiant is visually detected at a small offset of 46 mas and if gravitationally bound to the SB system would have an orbital period of many decades. Theta Mus is X-ray bright and a nonthermal radio source as commonly observed in massive colliding wind (CW) binaries. We present new Chandra X-ray observations of Theta Mus which complement previous XMM-Newton observations. The X-ray emission consists of a cool nearly steady weakly-absorbed plasma component with broad redshifted emission lines located in an extended region far from the SB system. Hotter plasma is also present traced by Fe XXV emission. The observed flux in the 2-5 keV range dropped significantly on a timescale of less than 5 years. The flux decrease can be attributed to an increase in absorption toward the hotter plasma which is likely located in the confined wind interaction region of the short-period SB system. The X-ray emission of Theta Mus is remarkably similar to the WC+O binary gamma^2 Vel including carbon recombination spectral lines but both systems show unusual line centroid properties that challenge CW models.
|
astro-ph
|
2,312.04498 |
The search for strategies to harness the temperature of quantum systems is one of the main goals in quantum thermodynamics. Here we study the dynamics of a system made of a pair of quantum harmonic oscillators, represented by single-mode cavity fields, interacting with a thermally excited beam of phaseonium atoms, which act as ancillas. The two cavities are arranged in a cascade configuration, so that the second cavity interacts with phaseonium atoms only after their interaction with the first one. We provide exact closed dynamics of the first cavity for arbitrarily long interaction times. We highlight the role played by the characteristic coherence phase of phaseonium atoms in determining the steady states of the cavity fields as well as that of the ancillas. Also, we show how the second cavity follows a non-Markovian evolution due to interactions with the "used" ancillary atoms, that enables information exchange with the first cavity. Adjusting the parameters of the phaseonium atoms, we can determine the final stable temperature reached by the cavities. In this way, the cavities can be heated up as well as cooled down. These results provide useful insights towards the use of different types of ancillas for thermodynamic cycles in cavity QED scenarios.
|
quant-ph
|
2,312.02401 |
Content moderation at scale faces the challenge of considering local cultural distinctions when assessing content. While global policies aim to maintain decision-making consistency and prevent arbitrary rule enforcement, they often overlook regional variations in interpreting natural language as expressed in content. In this study, we are looking into how moderation systems can tackle this issue by adapting to local comprehension nuances. We train large language models on extensive datasets of media news and articles to create culturally attuned models. The latter aim to capture the nuances of communication across geographies with the goal of recognizing cultural and societal variations in what is considered offensive content. We further explore the capability of these models to generate explanations for instances of content violation, aiming to shed light on how policy guidelines are perceived when cultural and societal contexts change. We find that training on extensive media datasets successfully induced cultural awareness and resulted in improvements in handling content violations on a regional basis. Additionally, these advancements include the ability to provide explanations that align with the specific local norms and nuances as evidenced by the annotators' preference in our conducted study. This multifaceted success reinforces the critical role of an adaptable content moderation approach in keeping pace with the ever-evolving nature of the content it oversees.
|
cs
|
2,312.03239 |
In this paper, we discuss the asymptotic behaviour of the weak solution to the Cauchy problem for the scalar viscous conservation law, with nonlinear Laplacian viscosity. Firstly, we obtain the existence, uniqueness and regularity of solutions when the initial data $u_0\in C^1(\mathbb R^N)\cap W^{1,\infty}(\mathbb R^N)$. Secondly, when $u_0$ is periodic, we prove the time-decay rate of the periodic solution and its gradient. At last, we study the long-time behaviour of perturbed solution to the Cauchy problem, in which the initial data is a $N-$d periodic perturbation around a planar rarefaction wave and obtain the time-decay rate of the perturbed solution approaching approximate planar rarefaction wave. The proof is given by technical energy methods and iteration technique.
|
math
|
2,312.03079 |
We present LooseControl to allow generalized depth conditioning for diffusion-based image generation. ControlNet, the SOTA for depth-conditioned image generation, produces remarkable results but relies on having access to detailed depth maps for guidance. Creating such exact depth maps, in many scenarios, is challenging. This paper introduces a generalized version of depth conditioning that enables many new content-creation workflows. Specifically, we allow (C1) scene boundary control for loosely specifying scenes with only boundary conditions, and (C2) 3D box control for specifying layout locations of the target objects rather than the exact shape and appearance of the objects. Using LooseControl, along with text guidance, users can create complex environments (e.g., rooms, street views, etc.) by specifying only scene boundaries and locations of primary objects. Further, we provide two editing mechanisms to refine the results: (E1) 3D box editing enables the user to refine images by changing, adding, or removing boxes while freezing the style of the image. This yields minimal changes apart from changes induced by the edited boxes. (E2) Attribute editing proposes possible editing directions to change one particular aspect of the scene, such as the overall object density or a particular object. Extensive tests and comparisons with baselines demonstrate the generality of our method. We believe that LooseControl can become an important design tool for easily creating complex environments and be extended to other forms of guidance channels. Code and more information are available at .
|
cs
|
2,312.04313 |
Steering is a type of quantum nonlocality that exhibits an inherent asymmetry between two observers. In a nondegenerate three-level laser coupled to a two-mode squeezed vacuum reservoir, we examine, under realistic experimental conditions, the Gaussian steering of two laser modes, $\mathcal{A}$ and $\mathcal{B}$, generated within the cascade transitions, respectively. We find that the $\mathcal{A}\rightarrow \mathcal{B}$ steerability is always higher than that from $\mathcal{B}\rightarrow \mathcal{A}$; in addition, the steering asymmetry cannot exceed $\ln 2$, which implies that the state $\hat{\varrho}_{\mathcal{AB}}$ never diverges to an extremal asymmetry state. We show how squeezed noise can play a constructive role in realizing one-way steering. As the main result, we demonstrate that the state $\hat{\varrho}_{\mathcal{AB}}$ can exhibit one-way steering solely from $\mathcal{A}\rightarrow \mathcal{B}$, which we show to emerge as a consequence of the fact that the intensity difference of the modes $\mathcal{A}$ and $\mathcal{B}$ is verified to remain always positive, irrespective of the physical and environmental parameters of $\hat{\varrho}_{\mathcal{AB}}$. The generated unidirectional one-way steering may provide a useful resource for the distribution of the trust in future asymmetric quantum information tasks.
|
quant-ph
|
2,312.02473 |
Existing Graph Neural Network (GNN) training frameworks have been designed to help developers easily create performant GNN implementations. However, most existing GNN frameworks assume that the input graphs are static, but ignore that most real-world graphs are constantly evolving. Though many dynamic GNN models have emerged to learn from evolving graphs, the training process of these dynamic GNNs is dramatically different from traditional GNNs in that it captures both the spatial and temporal dependencies of graph updates. This poses new challenges for designing dynamic GNN training frameworks. First, the traditional batched training method fails to capture real-time structural evolution information. Second, the time-dependent nature makes parallel training hard to design. Third, it lacks system supports for users to efficiently implement dynamic GNNs. In this paper, we present NeutronStream, a framework for training dynamic GNN models. NeutronStream abstracts the input dynamic graph into a chronologically updated stream of events and processes the stream with an optimized sliding window to incrementally capture the spatial-temporal dependencies of events. Furthermore, NeutronStream provides a parallel execution engine to tackle the sequential event processing challenge to achieve high performance. NeutronStream also integrates a built-in graph storage structure that supports dynamic updates and provides a set of easy-to-use APIs that allow users to express their dynamic GNNs. Our experimental results demonstrate that, compared to state-of-the-art dynamic GNN implementations, NeutronStream achieves speedups ranging from 1.48X to 5.87X and an average accuracy improvement of 3.97%.
|
cs
|
2,312.03345 |
6-DoF object-agnostic grasping in unstructured environments is a critical yet challenging task in robotics. Most current works use non-optimized approaches to sample grasp locations and learn spatial features without concerning the grasping task. This paper proposes GraNet, a graph-based grasp pose generation framework that translates a point cloud scene into multi-level graphs and propagates features through graph neural networks. By building graphs at the scene level, object level, and grasp point level, GraNet enhances feature embedding at multiple scales while progressively converging to the ideal grasping locations by learning. Our pipeline can thus characterize the spatial distribution of grasps in cluttered scenes, leading to a higher rate of effective grasping. Furthermore, we enhance the representation ability of scalable graph networks by a structure-aware attention mechanism to exploit local relations in graphs. Our method achieves state-of-the-art performance on the large-scale GraspNet-1Billion benchmark, especially in grasping unseen objects (+11.62 AP). The real robot experiment shows a high success rate in grasping scattered objects, verifying the effectiveness of the proposed approach in unstructured environments.
|
cs
|
2,312.04556 |
Large language models (LLMs) such as ChatGPT have received immense interest for their general-purpose language understanding and, in particular, their ability to generate high-quality text or computer code. For many professions, LLMs represent an invaluable tool that can speed up and improve the quality of work. In this note, we discuss to what extent they can aid professional mathematicians. We first provide a mathematical description of the transformer model used in all modern language models. Based on recent studies, we then outline best practices and potential issues and report on the mathematical abilities of language models. Finally, we shed light on the potential of LMMs to change how mathematicians work.
|
cs
|
2,312.03876 |
Weather forecasting is a fundamental problem for anticipating and mitigating the impacts of climate change. Recently, data-driven approaches for weather forecasting based on deep learning have shown great promise, achieving accuracies that are competitive with operational systems. However, those methods often employ complex, customized architectures without sufficient ablation analysis, making it difficult to understand what truly contributes to their success. Here we introduce Stormer, a simple transformer model that achieves state-of-the-art performance on weather forecasting with minimal changes to the standard transformer backbone. We identify the key components of Stormer through careful empirical analyses, including weather-specific embedding, randomized dynamics forecast, and pressure-weighted loss. At the core of Stormer is a randomized forecasting objective that trains the model to forecast the weather dynamics over varying time intervals. During inference, this allows us to produce multiple forecasts for a target lead time and combine them to obtain better forecast accuracy. On WeatherBench 2, Stormer performs competitively at short to medium-range forecasts and outperforms current methods beyond 7 days, while requiring orders-of-magnitude less training data and compute. Additionally, we demonstrate Stormer's favorable scaling properties, showing consistent improvements in forecast accuracy with increases in model size and training tokens. Code and checkpoints will be made publicly available.
|
cs
|
2,312.05948 |
This study presents an investigation into the thermodynamic properties of a dirty black hole immersed in a uniform electric field within the framework of the Einstein-Nonlinear Electrodynamics (ENE)-dilaton theory. The analysis delves into various thermodynamic aspects, including heat capacity, Helmholtz free energy, and internal energy, providing insights into the behavior of the black hole under the influence of the electric field. Furthermore, the article explores the intricate interplay between quantum effects and thermodynamic behavior through the examination of quantum-corrected entropy. The study aims to shed light on the non-perturbative corrections that arise in this complex system, offering a comprehensive understanding of the modified thermodynamics of dirty black holes within the specified theoretical framework.
|
quant-ph
|
2,312.05623 |
Prior works have analyzed the performance of millimeter wave automotive radars in the presence of diverse clutter and interference scenarios using stochastic geometry tools instead of more time-consuming measurement studies or system-level simulations. In these works, the distributions of radars or discrete clutter scatterers were modeled as Poisson point processes in the Euclidean space. However, since most automotive radars are likely to be mounted on vehicles and road infrastructure, road geometries are an important factor that must be considered. Instead of considering each road geometry as an individual case for study, in this work, we model each case as a specific instance of an underlying Poisson line process and further model the distribution of vehicles on the road as a Poisson point process - forming a Poisson line Cox process. Then, through the use of stochastic geometry tools, we estimate the average number of interfering radars for specific road and vehicular densities and the effect of radar parameters such as noise and beamwidth on the radar detection metrics. The numerical results are validated with Monte Carlo simulations.
|
math
|
2,312.04307 |
In active learning for graph-structured data, Graph Neural Networks (GNNs) have shown effectiveness. However, a common challenge in these applications is the underutilization of crucial structural information. To address this problem, we propose the Structural-Clustering PageRank method for improved Active learning (SPA) specifically designed for graph-structured data. SPA integrates community detection using the SCAN algorithm with the PageRank scoring method for efficient and informative sample selection. SPA prioritizes nodes that are not only informative but also central in structure. Through extensive experiments, SPA demonstrates higher accuracy and macro-F1 score over existing methods across different annotation budgets and achieves significant reductions in query time. In addition, the proposed method only adds two hyperparameters, $\epsilon$ and $\mu$ in the algorithm to finely tune the balance between structural learning and node selection. This simplicity is a key advantage in active learning scenarios, where extensive hyperparameter tuning is often impractical.
|
cs
|
2,312.04967 |
Control theory deals with the study of controlling dynamical systems. Robots today are growing increasingly complex and moving out of factory floors to real world environment. These robots have to interact with real world environment factors such as disturbances and this requires the robot to have a control system that is robust. Testing control algorithms on robots in real world environment can pose critical safety issues and can be financially expensive. This has resulted in a heavy emphasis on using simulation to test control algorithms before deploying them in real world environments. Designing control algorithms is an iterative process that starts with modelling the target system in simulation, designing a controller, testing the controller in simulation and then changing the controller parameters to design a better controller. This report explores how an approximated system model of a target hardware system can be developed, which can then be used to design a LQR controller for the target system. The controller is then tested under a disturbance, on hardware and in simulation, and the system response is recorded. The system response from hardware and simulation are then compared to validate the use of approximated system models in simulation for designing and testing control algorithms.
|
cs
|
2,312.04087 |
With recent advancements in Large Multimodal Models (LMMs) across various domains, a novel prompting method called visual referring prompting has emerged, showing significant potential in enhancing human-computer interaction within multimodal systems. This method offers a more natural and flexible approach to human interaction with these systems compared to traditional text descriptions or coordinates. However, the categorization of visual referring prompting remains undefined, and its impact on the performance of LMMs has yet to be formally examined. In this study, we conduct the first comprehensive analysis of LMMs using a variety of visual referring prompting strategies. We introduce a benchmark dataset called VRPTEST, comprising 3 different visual tasks and 2,275 images, spanning diverse combinations of prompt strategies. Using VRPTEST, we conduct a comprehensive evaluation of eight versions of prominent open-source and proprietary foundation models, including two early versions of GPT-4V. We develop an automated assessment framework based on software metamorphic testing techniques to evaluate the accuracy of LMMs without the need for human intervention or manual labeling. We find that the current proprietary models generally outperform the open-source ones, showing an average accuracy improvement of 22.70%; however, there is still potential for improvement. Moreover, our quantitative analysis shows that the choice of prompt strategy significantly affects the accuracy of LMMs, with variations ranging from -17.5% to +7.3%. Further case studies indicate that an appropriate visual referring prompting strategy can improve LMMs' understanding of context and location information, while an unsuitable one might lead to answer rejection. We also provide insights on minimizing the negative impact of visual referring prompting on LMMs.
|
cs
|
2,312.0572 |
Federated learning (FL) emphasizes decentralized training by storing data locally and sending only model updates, underlining user privacy. Recently, a line of works on privacy attacks impairs user privacy by extracting sensitive training text from language models in the context of FL. Yet, these attack techniques face distinct hurdles: some work chiefly with limited batch sizes (e.g., batch size of 1), and others are easily detectable. This paper introduces an innovative approach that is challenging to detect, significantly enhancing the recovery rate of text in various batch-size settings. Building on fundamental gradient matching and domain prior knowledge, we enhance the attack by recovering the input of the Pooler layer of language models, which enables us to provide additional supervised signals at the feature level. Unlike gradient data, these signals do not average across sentences and tokens, thereby offering more nuanced and effective insights. We benchmark our method using text classification tasks on datasets such as CoLA, SST-2, and Rotten Tomatoes. Across different batch sizes and models, our approach consistently outperforms previous state-of-the-art results.
|
cs
|
2,312.05121 |
We prove that the kissing number in 48 dimensions among antipodal spherical codes with certain forbidden inner products is 52\,416\,000. Constructions of attaining codes as kissing configurations of minimum vectors in even unimodular extremal lattices are well known since the 1970's. We also prove that corresponding spherical 11-designs with the same cardinality are minimal. We use appropriate modifications of the linear programming bounds for spherical codes and designs introduced by Delsarte, Goethals and Seidel in 1977.
|
math
|
2,312.02538 |
Dense retrieval methods have been mostly focused on unstructured text and less attention has been drawn to structured data with various aspects, e.g., products with aspects such as category and brand. Recent work has proposed two approaches to incorporate the aspect information into item representations for effective retrieval by predicting the values associated with the item aspects. Despite their efficacy, they treat the values as isolated classes (e.g., "Smart Homes", "Home, Garden & Tools", and "Beauty & Health") and ignore their fine-grained semantic relation. Furthermore, they either enforce the learning of aspects into the CLS token, which could confuse it from its designated use for representing the entire content semantics, or learn extra aspect embeddings only with the value prediction objective, which could be insufficient especially when there are no annotated values for an item aspect. Aware of these limitations, we propose a MUlti-granulaRity-aware Aspect Learning model (MURAL) for multi-aspect dense retrieval. It leverages aspect information across various granularities to capture both coarse and fine-grained semantic relations between values. Moreover, MURAL incorporates separate aspect embeddings as input to transformer encoders so that the masked language model objective can assist implicit aspect learning even without aspect-value annotations. Extensive experiments on two real-world datasets of products and mini-programs show that MURAL outperforms state-of-the-art baselines significantly.
|
cs
|
2,312.04437 |
We investigate the impact of measuring one subsystem on the holographic complexity of another. While a naive expectation might suggest a reduction in complexity due to the collapse of the state to a trivial product state during quantum measurements, our findings reveal a counterintuitive result: in numerous scenarios, measurements on one subsystem can amplify the complexity of another. We first present a counting argument elucidating this complexity transition in random states. Then, employing the subregion "complexity=volume" (CV) proposal, we identify a complexity phase transition induced by projection measurements in various holographic CFT setups, including CFT vacuum states, thermofield double states, and the joint system of a black hole coupled to a bath. According to the AdS/BCFT correspondence, the post-measurement dual geometry involves an end-of-the-world brane created by the projection measurement. The complexity phase transition corresponds to the transition of the entanglement wedge to the one connected to the brane. In the context of the thermofield double setup, complete projection on one side can transform the other side into a boundary state black hole with higher complexity or a pure AdS with lower complexity. In the joint system of a black hole coupled to a nongraviting bath, where (a part of) the radiation is measured, the BCFT features two boundaries: one for the black hole and the other for the measurement. We construct the bulk dual involving intersecting or non-intersecting branes, and investigate the complexity transition induced by the projection measurement. Notably, for a subsystem that contains the black hole brane, its RT surface may undergo a transition, giving rise to a complexity jump.
|
quant-ph
|
2,312.04484 |
LiDAR segmentation is crucial for autonomous driving systems. The recent range-view approaches are promising for real-time processing. However, they suffer inevitably from corrupted contextual information and rely heavily on post-processing techniques for prediction refinement. In this work, we propose a simple yet powerful FRNet that restores the contextual information of the range image pixels with corresponding frustum LiDAR points. Firstly, a frustum feature encoder module is used to extract per-point features within the frustum region, which preserves scene consistency and is crucial for point-level predictions. Next, a frustum-point fusion module is introduced to update per-point features hierarchically, which enables each point to extract more surrounding information via the frustum features. Finally, a head fusion module is used to fuse features at different levels for final semantic prediction. Extensive experiments on four popular LiDAR segmentation benchmarks under various task setups demonstrate our superiority. FRNet achieves competitive performance while maintaining high efficiency. The code is publicly available.
|
cs
|
2,312.06147 |
Large language models (LLMs) that have been trained on a corpus that includes large amount of code exhibit a remarkable ability to understand HTML code. As web interfaces are primarily constructed using HTML, we design an in-depth study to see how LLMs can be used to retrieve and locate important elements for a user given query (i.e. task description) in a web interface. In contrast with prior works, which primarily focused on autonomous web navigation, we decompose the problem as an even atomic operation - Can LLMs identify the important information in the web page for a user given query? This decomposition enables us to scrutinize the current capabilities of LLMs and uncover the opportunities and challenges they present. Our empirical experiments show that while LLMs exhibit a reasonable level of performance in retrieving important UI elements, there is still a substantial room for improvement. We hope our investigation will inspire follow-up works in overcoming the current challenges in this domain.
|
cs
|
2,312.05234 |
There are various applications, where companies need to decide to which individuals they should best allocate treatment. To support such decisions, uplift models are applied to predict treatment effects on an individual level. Based on the predicted treatment effects, individuals can be ranked and treatment allocation can be prioritized according to this ranking. An implicit assumption, which has not been doubted in the previous uplift modeling literature, is that this treatment prioritization approach tends to bring individuals with high treatment effects to the top and individuals with low treatment effects to the bottom of the ranking. In our research, we show that heteroskedastictity in the training data can cause a bias of the uplift model ranking: individuals with the highest treatment effects can get accumulated in large numbers at the bottom of the ranking. We explain theoretically how heteroskedasticity can bias the ranking of uplift models and show this process in a simulation and on real-world data. We argue that this problem of ranking bias due to heteroskedasticity might occur in many real-world applications and requires modification of the treatment prioritization to achieve an efficient treatment allocation.
|
cs
|
2,312.03413 |
The Knapsack Problem is a classic problem in combinatorial optimisation. Solving these problems may be computationally expensive. Recent years have seen a growing interest in the use of deep learning methods to approximate the solutions to such problems. A core problem is how to enforce or encourage constraint satisfaction in predicted solutions. A promising approach for predicting solutions to constrained optimisation problems is the Lagrangian Dual Framework which builds on the method of Lagrangian Relaxation. In this paper we develop neural network models to approximate Knapsack Problem solutions using the Lagrangian Dual Framework while improving constraint satisfaction. We explore the problems of output interpretation and model selection within this context. Experimental results show strong constraint satisfaction with a minor reduction of optimality as compared to a baseline neural network which does not explicitly model the constraints.
|
math
|
2,312.03722 |
Carbon accounting is a fundamental building block in our global path to emissions reduction and decarbonization, yet many challenges exist in achieving reliable and trusted carbon accounting measures. We motivate that carbon accounting not only needs to be more data-driven, but also more methodologically sound. We discuss the need for alternative, more diverse data sources that can play a significant role on our path to trusted carbon accounting procedures and elaborate on not only why, but how Artificial Intelligence (AI) in general and Natural Language Processing (NLP) in particular can unlock reasonable access to a treasure trove of alternative data sets in light of the recent advances in the field that better enable the utilization of unstructured data in this process. We present a case study of the recent developments on real-world data via an NLP-powered analysis using OpenAI's GPT API on financial and shipping data. We conclude the paper with a discussion on how these methods and approaches can be integrated into a broader framework for AI-enabled integrative carbon accounting.
|
cs
|
2,312.03476 |
We present a novel method for measuring the lags of (weak) variability components in neutron-star and black-hole low-mass X-ray binaries (LMXBs). For this we assume that the power and cross spectra of these sources consists of a number of components that are coherent in different energy bands, but are incoherent with one another. The technique is based on fitting simultaneously the power spectrum (PS) and the Real and Imaginary parts of the cross spectrum (CS) with a combination of Lorentzian functions. We show that, because the PS of LMXBs is insensitive to signals with a large Imaginary part and a small Real part in the CS, this approach allows us to uncover new variability components that are only detected in the CS. We also demonstrate that, contrary to earlier claims, the frequency of the type-C quasi-periodic oscillation (QPO) in the black-hole binary GRS 1915+105 does not depend on energy. Rather, the apparent energy dependence of the QPO frequency can be explained by the presence of a separate QPO component with a slightly higher frequency than that of the QPO, whose rms amplitude increases faster with energy than the rms amplitude of the QPO. From all the above we conclude that, as in the case of the PS, the CS of black-hole and neutron-star binaries can be fitted by a combination of Lorentzian components. Our findings provide evidence that the frequency-dependent part of the transfer function of these systems can be described by a combination of responses, each of them acting over relatively well-defined time scales. This conclusion challenges models that assume that the main contribution to the lags comes from a global, broadband, transfer function of the accreting system.
|
astro-ph
|
2,312.05428 |
Formation coordination is a critical aspect of swarm robotics, which involves coordinating the motion and behavior of a group of robots to achieve a specific objective. In formation coordination, the robots must maintain a specific spatial arrangement while in motion. In this paper, we present a leader-follower column formation coordination problem in an unknown, two-dimensional nonlinear manifold, where we redefining it as a trajectory estimation problem. Leveraging Koopman operator theory and Extended Dynamic Mode Decomposition, we estimate the measurement vectors for the follower agent and guide its nonlinear trajectories.
|
eess
|
2,312.04742 |
Modern communication systems need to fulfill multiple and often conflicting objectives at the same time. In particular, new applications require high reliability while operating at low transmit powers. Moreover, reliability constraints may vary over time depending on the current state of the system. One solution to address this problem is to use joint transmissions from a number of base stations (BSs) to meet the reliability requirements. However, this approach is inefficient when considering the overall total transmit power. In this work, we propose a reinforcement learning-based power allocation scheme for an unmanned aerial vehicle (UAV) communication system with varying communication reliability requirements. In particular, the proposed scheme aims to minimize the total transmit power of all BSs while achieving an outage probability that is less than a tolerated threshold. This threshold varies over time, e.g., when the UAV enters a critical zone with high-reliability requirements. Our results show that the proposed learning scheme uses dynamic power allocation to meet varying reliability requirements, thus effectively conserving energy.
|
eess
|
2,312.03535 |
Let $\Phi$ be a pseudo-Anosov diffeomorphism of a compact (possibily non-orientable) surface $\Sigma$ with one boundary component. We show that if $b \in \pi_1(\Sigma)$ is the boundary word, $\phi \in {\rm{Aut}}(\pi_1(\Sigma))$ is a representative of $\Phi$ fixing $b$, and ${\rm{ad}}_b$ denotes conjugation by $b$, then the orbits of $\langle \phi, {\rm{ad}}_b \rangle\cong\mathbb{Z}^2$ in the graph of free factors of $\pi_1(\Sigma)$ are quasi-isometrically embedded. It follows that for $N \geq 2$ the free factor graph for ${\rm{Aut}}(F_N)$ is not hyperbolic, in contrast to the ${\rm{Out}}(F_N)$ case.
|
math
|
2,312.04679 |
tmospheric turbulence presents a significant challenge in long-range imaging. Current restoration algorithms often struggle with temporal inconsistency, as well as limited generalization ability across varying turbulence levels and scene content different than the training data. To tackle these issues, we introduce a self-supervised method, Consistent Video Restoration through Turbulence (ConVRT) a test-time optimization method featuring a neural video representation designed to enhance temporal consistency in restoration. A key innovation of ConVRT is the integration of a pretrained vision-language model (CLIP) for semantic-oriented supervision, which steers the restoration towards sharp, photorealistic images in the CLIP latent space. We further develop a principled selection strategy of text prompts, based on their statistical correlation with a perceptual metric. ConVRT's test-time optimization allows it to adapt to a wide range of real-world turbulence conditions, effectively leveraging the insights gained from pre-trained models on simulated data. ConVRT offers a comprehensive and effective solution for mitigating real-world turbulence in dynamic videos.
|
cs
|
2,312.04634 |
We present six spectroscopically confirmed massive protostructures, spanning a redshift range of $2.5<z<4.5$ in the Extended Chandra Deep Field South (ECDFS) field discovered as part of the Charting Cluster Construction in VUDS and ORELSE (C3VO) survey. We identify and characterize these remarkable systems by applying an overdensity measurement technique on an extensive data compilation of public and proprietary spectroscopic and photometric observations in this highly studied extragalactic field. Each of these six protostructures, i.e., a large scale overdensity (volume $>9000$\thinspace cMpc$^3$) of more than $2.5\sigma_{\delta}$ above the field density levels at these redshifts, have a total mass $M_{tot}\ge10^{14.8}M_\odot$ and one or more highly overdense (overdensity$\thinspace>5\sigma_{\delta}$) peaks. One of the most complex protostructures discovered is a massive ($M_{tot}=10^{15.1}M_\odot$) system at $z\sim3.47$ that contains six peaks and 55 spectroscopic members. We also discover protostructures at $z\sim3.30$ and $z\sim3.70$ that appear to at least partially overlap on sky with the protostructure at $z\sim3.47$, suggesting a possible connection. We additionally report on the discovery of three massive protostructures at $z=2.67$, 2.80, and 4.14 and discuss their properties. Finally, we discuss the relationship between star formation rate and environment in the richest of these protostructures, finding an enhancement of star formation activity in the densest regions. The diversity of the protostructures reported here provide an opportunity to study the complex effects of dense environments on galaxy evolution over a large redshift range in the early universe.
|
astro-ph
|
2,312.02521 |
Recent diffusion-based subject driven generative methods have enabled image generations with good fidelity for specific objects or human portraits. However, to achieve better versatility for applications, we argue that not only improved datasets and evaluations are desired, but also more careful methods to retrieve only relevant information from conditional images are anticipated. To this end, we propose an anime figures dataset RetriBooru-V1, with enhanced identity and clothing labels. We state new tasks enabled by this dataset, and introduce a new diversity metric to measure success in completing these tasks, quantifying the flexibility of image generations. We establish an RAG-inspired baseline method, designed to retrieve precise conditional information from reference images. Then, we compare with current methods on existing task to demonstrate the capability of the proposed method. Finally, we provide baseline experiment results on new tasks, and conduct ablation studies on the possible structural choices.
|
cs
|
2,312.04457 |
Let $X$ be a chemical reaction process, modeled as a multi-dimensional continuous-time jump process. Assume that at given times $0<t_1 < \cdots < t_n$, linear combinations $v_i = L_i X(t_i),\, i = 1,\dots, n$ are observed for given matrices $L_i$. We show how to sample the process conditioned on hitting the states $v_1,\dots, v_n$ by a change of measure on the law of the unconditioned process. We derive sufficient conditions for this change of measure and complement our results using numerical illustrations.
|
math
|
2,312.03728 |
Large Language Models (LLMs), as the case of OpenAI ChatGPT-4 Turbo, are revolutionizing several industries, including higher education. In this context, LLMs can be personalized through a fine-tuning process to meet the student demands on every particular subject, like statistics. Recently, OpenAI has launched the possibility to fine-tune their model with a natural language web interface, enabling the possibility to create customized GPT version deliberately conditioned to meet the demands of a specific task. The objective of this research is to assess the potential of the customized GPTs that have recently been launched by OpenAI. After developing a Business Statistics Virtual Professor (BSVP), tailored for students at the Universidad Pontificia Comillas, its behavior was evaluated and compared with that of ChatGPT-4 Turbo. The results lead to several conclusions. Firstly, a substantial modification in the style of communication was observed. Following the instructions it was trained with, BSVP provided responses in a more relatable and friendly tone, even incorporating a few minor jokes. Secondly, and this is a matter of relevance, when explicitly asked for something like, "I would like to practice a programming exercise similar to those in R practice 4," BSVP was capable of providing a far superior response: having access to contextual documentation, it could fulfill the request, something beyond ChatGPT-4 Turbo's capabilities. On the downside, the response times were generally higher. Lastly, regarding overall performance, quality, depth, and alignment with the specific content of the course, no statistically significant differences were observed in the responses between BSVP and ChatGPT-4 Turbo. It appears that customized assistants trained with prompts present advantages as virtual aids for students, yet they do not constitute a substantial improvement over ChatGPT-4 Turbo.
|
cs
|
2,312.0437 |
Diffusion models are a new class of generative models that have shown outstanding performance in image generation literature. As a consequence, studies have attempted to apply diffusion models to other tasks, such as speech enhancement. A popular approach in adapting diffusion models to speech enhancement consists in modelling a progressive transformation between the clean and noisy speech signals. However, one popular diffusion model framework previously laid in image generation literature did not account for such a transformation towards the system input, which prevents from relating the existing diffusion-based speech enhancement systems with the aforementioned diffusion model framework. To address this, we extend this framework to account for the progressive transformation between the clean and noisy speech signals. This allows us to apply recent developments from image generation literature, and to systematically investigate design aspects of diffusion models that remain largely unexplored for speech enhancement, such as the neural network preconditioning, the training loss weighting, the stochastic differential equation (SDE), or the amount of stochasticity injected in the reverse process. We show that the performance of previous diffusion-based speech enhancement systems cannot be attributed to the progressive transformation between the clean and noisy speech signals. Moreover, we show that a proper choice of preconditioning, training loss weighting, SDE and sampler allows to outperform a popular diffusion-based speech enhancement system in terms of perceptual metrics while using fewer sampling steps, thus reducing the computational cost by a factor of four.
|
cs
|
2,312.02635 |
Research on Multi-rotor Aerial Vehicles (MAVs) has experienced remarkable advancements over the past two decades, propelling the field forward at an accelerated pace. Through the implementation of motion control and the integration of specialized mechanisms, researchers have unlocked the potential of MAVs to perform a wide range of tasks in diverse scenarios. Notably, the literature has highlighted the distinctive attributes of MAVs that endow them with a competitive edge in physical interaction when compared to other robotic systems. In this survey, we present a categorization of the various types of physical interactions in which MAVs are involved, supported by comprehensive case studies. We examine the approaches employed by researchers to address different challenges using MAVs and their applications, including the development of different types of controllers to handle uncertainties inherent in these interactions. By conducting a thorough analysis of the strengths and limitations associated with different methodologies, as well as engaging in discussions about potential enhancements, this survey aims to illuminate the path for future research focusing on MAVs with high actuation capabilities.
|
cs
|
2,312.04926 |
In recent years, pruning has emerged as a popular technique to reduce the computational complexity and memory footprint of Convolutional Neural Network (CNN) models. Mutual Information (MI) has been widely used as a criterion for identifying unimportant filters to prune. However, existing methods for MI computation suffer from high computational cost and sensitivity to noise, leading to suboptimal pruning performance. We propose a novel method to improve MI computation for CNN pruning, using the spatial aura entropy. The spatial aura entropy is useful for evaluating the heterogeneity in the distribution of the neural activations over a neighborhood, providing information about local features. Our method effectively improves the MI computation for CNN pruning, leading to more robust and efficient pruning. Experimental results on the CIFAR-10 benchmark dataset demonstrate the superiority of our approach in terms of pruning performance and computational efficiency.
|
cs
|
2,312.04612 |
Let $\Phi'$ denote the strong dual of a nuclear space $\Phi$ and let $C_{\infty}(\Phi')$ be the collection of all continuous mappings $x:[0,\infty) \rightarrow \Phi'$ equipped with the topology of local uniform convergence. In this paper we introduce sufficient conditions for tightness of probability measures on $C_{\infty}(\Phi')$ and for weak convergence on $C_{\infty}(\Phi')$ for a sequence of $\Phi'$-valued processes.
|
math
|
2,312.03663 |
In $H$-percolation, we start with an Erdős--Rényi graph ${\mathcal G}_{n,p}$ and then iteratively add edges that complete copies of $H$. The process percolates if all edges missing from ${\mathcal G}_{n,p}$ are eventually added. We find the critical threshold $p_c$ when $H={\mathcal G}_{k,1/2}$ is uniformly random, solving a problem of Balogh, Bollobás and Morris.
|
math
|
2,312.02971 |
Row-column multiplexing has proven to be an effective strategy in scaling single-photon detector arrays to kilopixel and megapixel spatial resolutions. However, with this readout mechanism, multiphoton coincidences on the array cannot be easily resolved due to ambiguities concerning their spatial locations of incidence. In this work, we propose a method to resolve up to 4-photon coincidences in single-photon detector arrays with row-column readouts. By utilizing unambiguous single-photon measurements to estimate probabilities of detection at each pixel, we redistribute the ambiguous multiphoton counts among candidate pixel locations such that the peak signal-to-noise-ratio of the reconstruction is increased between 3 and 4 dB compared to conventional methods at optimal operating conditions. We also show that our method allows the operation of these arrays at higher incident photon fluxes as compared to previous methods. The application of this technique to imaging natural scenes is demonstrated using Monte Carlo experiments.
|
quant-ph
|
2,312.06203 |
AI-Generated Content (AIGC), as a novel manner of providing Metaverse services in the forthcoming Internet paradigm, can resolve the obstacles of immersion requirements. Concurrently, edge computing, as an evolutionary paradigm of computing in communication systems, effectively augments real-time interactive services. In pursuit of enhancing the accessibility of AIGC services, the deployment of AIGC models (e.g., diffusion models) to edge servers and local devices has become a prevailing trend. Nevertheless, this approach faces constraints imposed by battery life and computational resources when tasks are offloaded to local devices, limiting the capacity to deliver high-quality content to users while adhering to stringent latency requirements. So there will be a tradeoff between the utility of AIGC models and offloading decisions in the edge computing paradigm. This paper proposes a joint optimization algorithm for offloading decisions, computation time, and diffusion steps of the diffusion models in the reverse diffusion stage. Moreover, we take the average error into consideration as the metric for evaluating the quality of the generated results. Experimental results conclusively demonstrate that the proposed algorithm achieves superior joint optimization performance compared to the baselines.
|
cs
|
2,312.03146 |
In-memory computing (IMC) with non-volatile memories (NVMs) has emerged as a promising approach to address the rapidly growing computational demands of Deep Neural Networks (DNNs). Mapping DNN layers spatially onto NVM-based IMC accelerators achieves high degrees of parallelism. However, two challenges that arise in this approach are the highly non-uniform distribution of layer processing times and high area requirements. We propose LRMP, a method to jointly apply layer replication and mixed precision quantization to improve the performance of DNNs when mapped to area-constrained NVM-based IMC accelerators. LRMP uses a combination of reinforcement learning and integer linear programming to search the replication-quantization design space using a model that is closely informed by the target hardware architecture. Across five DNN benchmarks, LRMP achieves 2.8-9$\times$ latency and 11.8-19$\times$ throughput improvement at iso-accuracy.
|
cs
|
2,312.05771 |
Meta-learning enables rapid generalization to new tasks by learning meta-knowledge from a variety of tasks. It is intuitively assumed that the more tasks a model learns in one training batch, the richer knowledge it acquires, leading to better generalization performance. However, contrary to this intuition, our experiments reveal an unexpected result: adding more tasks within a single batch actually degrades the generalization performance. To explain this unexpected phenomenon, we conduct a Structural Causal Model (SCM) for causal analysis. Our investigation uncovers the presence of spurious correlations between task-specific causal factors and labels in meta-learning. Furthermore, the confounding factors differ across different batches. We refer to these confounding factors as ``Task Confounders". Based on this insight, we propose a plug-and-play Meta-learning Causal Representation Learner (MetaCRL) to eliminate task confounders. It encodes decoupled causal factors from multiple tasks and utilizes an invariant-based bi-level optimization mechanism to ensure their causality for meta-learning. Extensive experiments on various benchmark datasets demonstrate that our work achieves state-of-the-art (SOTA) performance.
|
stat
|
2,312.0372 |
Large language models LLMs like ChatGPT have reached the 100 Mio user barrier in record time and might increasingly enter all areas of our life leading to a diverse set of interactions between those Artificial Intelligence models and humans. While many studies have discussed governance and regulations deductively from first-order principles, few studies provide an inductive, data-driven lens based on observing dialogues between humans and LLMs especially when it comes to non-collaborative, competitive situations that have the potential to pose a serious threat to people. In this work, we conduct a user study engaging over 40 individuals across all age groups in price negotiations with an LLM. We explore how people interact with an LLM, investigating differences in negotiation outcomes and strategies. Furthermore, we highlight shortcomings of LLMs with respect to their reasoning capabilities and, in turn, susceptiveness to prompt hacking, which intends to manipulate the LLM to make agreements that are against its instructions or beyond any rationality. We also show that the negotiated prices humans manage to achieve span a broad range, which points to a literacy gap in effectively interacting with LLMs.
|
cs
|
2,312.03188 |
Port-based teleportation (PBT) is a variant of quantum teleportation that, unlike the canonical protocol by Bennett et al., does not require a correction operation on the teleported state. Since its introduction by Ishizaka and Hiroshima in 2008, no efficient implementation of PBT was known. We close this long-standing gap by building on our recent results on representations of partially transposed permutation matrix algebras and mixed quantum Schur transform. We describe efficient quantum circuits for probabilistic and deterministic PBT protocols on $n$ ports of arbitrary local dimension, both for EPR and optimized resource states. We describe two constructions based on different encodings of the Gelfand-Tsetlin basis for $n$ qudits: a standard encoding that achieves $\widetilde{O}(n)$ time and $O(n\mathrm{log}(n))$ space complexity, and a Yamanouchi encoding that achieves $\widetilde{O}(n^2)$ time and $O(\mathrm{log}(n))$ space complexity, both for constant local dimension and target error. We also describe efficient circuits for preparing the optimal resource states.
|
math
|
2,312.04968 |
Known for their efficiency in analyzing large data sets, machine learning classifiers are widely used in wide-field sky surveys. The upcoming Vera C. Rubin Observatory Legacy of Time and Space Survey (LSST) will generate millions of alerts every night, enabling the discovery of large samples of rare events. Identifying such objects soon after explosion will be essential to study their evolution. This requires a machine learning framework that makes use of all available transient and contextual information. Using $\sim5400$ transients from the ZTF Bright Transient Survey as input data, we develop NEEDLE, a novel hybrid classifier to select for two rare classes with strong environmental preferences: superluminous supernovae (SLSNe) preferring dwarf galaxies, and tidal disruption events (TDEs) occurring in the centres of nucleated galaxies. The input data includes detection and reference images, photometric information from the alert packets, and host galaxy magnitudes from Pan-STARRS. Despite having only a few tens of examples of the rare classes, our average (best) completeness on an unseen test set reaches 77% (93%) for SLSNe and 72% (87%) for TDEs. This may still result in a large fraction of false positives for the rare transients, given the large class imbalance in real surveys. However, the goal of NEEDLE is to find good candidates for spectroscopic classification, rather than to select pure photometric samples. Our network is designed with LSST in mind and we expect performance to improve further with the higher resolution images and more accurate transient and host photometry that will be available from Rubin. Our system will be deployed as an annotator on the UK alert broker, Lasair, to provide predictions to the community in real time.
|
astro-ph
|
2,312.02181 |
With the rapid development of artificial intelligence and breakthroughs in machine learning and natural language processing, intelligent question-answering robots have become widely used in government affairs. This paper conducts a horizontal comparison between Guangdong Province's government chatbots, ChatGPT, and Wenxin Ernie, two large language models, to analyze the strengths and weaknesses of existing government chatbots and AIGC technology. The study finds significant differences between government chatbots and large language models. China's government chatbots are still in an exploratory stage and have a gap to close to achieve "intelligence." To explore the future direction of government chatbots more deeply, this research proposes targeted optimization paths to help generative AI be effectively applied in government chatbot conversations.
|
cs
|
2,312.03655 |
Let $E$ be an elliptic curve defined over a number field $k$ and $\ell$ a prime integer. When $E$ has at least one $k$-rational point of exact order $\ell$, we derive a uniform upper bound $\exp(C \log B / \log \log B)$ for the number of points of $E(k)$ of (exponential) height at most $B$. Here the constant $C = C(k)$ depends on the number field $k$ and is effective. For $\ell = 2$ this generalizes a result of Naccarato which applies for $k=\mathbb{Q}$. We follow methods previously developed by Bombieri and Zannier and further by Naccarato, with the main novelty being the application of Rosen's result on bounding $\ell$-ranks of class groups in certain extensions, which is derived using relative genus theory.
|
math
|
2,312.05022 |
We consider the following question arising in the theory of differential inclusions: given an elliptic set $\Gamma$ and a Sobolev map $u$ whose gradient lies in the quasiconformal envelope of $\Gamma$ and touches $\Gamma$ on a set of positive measure, must $u$ be affine? We answer this question positively for a suitable notion of ellipticity, which for instance encompasses the case where $\Gamma \subset \mathbb R^{2\times 2}$ is an elliptic, smooth, closed curve. More precisely, we prove that the distance of $D u$ to $\Gamma$ satisfies the strong unique continuation property. As a by-product, we obtain new results for nonlinear Beltrami equations and recover known results for the reduced Beltrami equation and the Monge--Ampère equation: concerning the latter, we obtain a new proof of the $W^{2,1+\varepsilon}$-regularity for two-dimensional solutions.
|
math
|
2,312.02685 |
Multivariate Bessel and Jacobi processes describe Calogero-Moser-Sutherland particle models. They depend on a parameter $k$ and are related to time-dependent classical random matrix models like Dysom Brownian motions, where $k$ has the interpretation of an inverse temperature. There are several stochastic limit theorems for $k\to\infty$ were the limits depend on the solutions of associated ODEs where these ODEs admit particular simple solutions which are connected with the zeros of the classical orthogonal polynomials. In this paper we show that these solutions attract all solutions. Moreover we present a connection between the solutions of these ODEs with associated inverse heat equations. These inverse heat equations are used to compute the expectations of some determinantal formulas for the Bessel and Jacobi processes.
|
math
|
2,312.05325 |
In this report, we delve into two critical research inquiries. Firstly, we explore the extent to which Reinforcement Learning (RL) agents exhibit multimodal distributions in the context of stop-and-go traffic scenarios. Secondly, we investigate how RL-controlled Robot Vehicles (RVs) effectively navigate their direction and coordinate with other vehicles in complex traffic environments. Our analysis encompasses an examination of multimodality within queue length, outflow, and platoon size distributions for both Robot and Human-driven Vehicles (HVs). Additionally, we assess the Pearson coefficient correlation, shedding light on relationships between queue length and outflow, considering both identical and differing travel directions. Furthermore, we delve into causal inference models, shedding light on the factors influencing queue length across scenarios involving varying travel directions. Through these investigations, this report contributes valuable insights into the behaviors of mixed traffic (RVs and HVs) in traffic management and coordination.
|
cs
|
2,312.02548 |
Data augmentation is crucial in training deep models, preventing them from overfitting to limited data. Common data augmentation methods are effective, but recent advancements in generative AI, such as diffusion models for image generation, enable more sophisticated augmentation techniques that produce data resembling natural images. We recognize that augmented samples closer to the ideal decision boundary of a classifier are particularly effective and efficient in guiding the learning process. We introduce GeNIe which leverages a diffusion model conditioned on a text prompt to merge contrasting data points (an image from the source category and a text prompt from the target category) to generate challenging samples for the target category. Inspired by recent image editing methods, we limit the number of diffusion iterations and the amount of noise. This ensures that the generated image retains low-level and contextual features from the source image, potentially conflicting with the target category. Our extensive experiments, in few-shot and also long-tail distribution settings, demonstrate the effectiveness of our novel augmentation method, especially benefiting categories with a limited number of examples.
|
cs
|
2,312.02782 |
In other to understand galaxy growth evolution, it is critical to constrain the evolution of its building block: gas. Mostly comprised by Hydrogen in its neutral (HI) and molecular (H$_2$) phases, the latter is the one mostly directly associated to star-formation, while the neutral phase is considered the long-term gas reservoir. In this work, we make use of an empirical relation between dust emission at millimeter wavelengths and total gas mass in the inter-stellar medium (MHI plus MH2) in order to retrieve the HI content in galaxies. We assemble an heterogeneous sample of 335 galaxies at $0.01<z<6.4$ detected in both mm-continuum and carbon monoxide (CO), with special focus on a blindly selected sample to retrieve HI cosmological content when the Universe was ~2-6Gyr old ($1<z<3$). We find no significant evolution with redshift of the $M_{\rm HI}/M_{\rm H_2}$ ratio, which is about 1-3 (depending on the relation used to estimate $M_{\rm HI}$). This also shows that $M_{\rm H_2}$-based gas depletion times are underestimated overall by a factor of 2-4. Compared to local Universe HI mass functions, we find that the number density of galaxies with $M_{\rm HI} > 10^{10.5} {\rm M}_\odot$ significantly decreased since 8-12Gyr ago. The specific sample used for this analysis is associated to 20-50% of the total cosmic HI content as estimated via Damped Lyman-alpha Absorbers. In IR luminous galaxies, HI mass content decreases between $z\sim 2.5$ and $z\sim 1.5$, while H$_2$ seems to increase. We also show source detection expectations for SKA surveys.
|
astro-ph
|
2,312.02948 |
We prove that the direct sums of extensions of scalars of relation modules are geometrically realisable as the second homotopy group of a finite 2-complex. We use this to exhibit a finite 2-complex with fundamental group the $(10,15)$ torus knot group and non-free $\pi_2$, yielding exotic presentations of a group for which no such examples had previously been known. We conclude by constructing stably free non-free modules over an infinite family of Baumslag-Solitar groups; it remains to determine whether these modules are geometrically realisable by finite 2-complexes.
|
math
|
2,312.06641 |
In many online sequential decision-making scenarios, a learner's choices affect not just their current costs but also the future ones. In this work, we look at one particular case of such a situation where the costs depend on the time average of past decisions over a history horizon. We first recast this problem with history dependent costs as a problem of decision making under stage-wise constraints. To tackle this, we then propose the novel Follow-The-Adaptively-Regularized-Leader (FTARL) algorithm. Our innovative algorithm incorporates adaptive regularizers that depend explicitly on past decisions, allowing us to enforce stage-wise constraints while simultaneously enabling us to establish tight regret bounds. We also discuss the implications of the length of history horizon on design of no-regret algorithms for our problem and present impossibility results when it is the full learning horizon.
|
cs
|
2,312.05301 |
We report near-infrared long-baseline interferometric observations of the Hyades multiple system HD 284163, made with the CHARA array, as well as almost 43 yr of high-resolution spectroscopic monitoring at the CfA. Both types of observations resolve the 2.39 d inner binary, and also an outer companion in a 43.1 yr orbit. Our observations, combined with others from the literature, allow us to solve for the 3D inner and outer orbits, which are found to be at nearly right angles to each other. We determine the dynamical masses of the three stars (good to better than 1.4% for the inner pair), as well as the orbital parallax. The secondary component (0.5245 +/- 0.0047 MSun) is now the lowest mass star with a dynamical mass measurement in the cluster. A comparison of these measurements with current stellar evolution models for the age and metallicity of the Hyades shows good agreement. All three stars display significant levels of chromospheric activity, consistent with the classification of HD 284163 as an RS CVn object. We present evidence that a more distant fourth star is physically associated, making this a hierarchical quadruple system.
|
astro-ph
|
2,312.03647 |
Double staining in histopathology, particularly for metaplastic breast cancer, typically employs H&E and P63 dyes. However, P63's tissue damage and high cost necessitate alternative methods. This study introduces xAI-CycleGAN, an advanced architecture combining Mask CycleGAN with explainability features and structure-preserving capabilities for transforming H&E stained breast tissue images into P63-like images. The architecture allows for output editing, enhancing resemblance to actual images and enabling further model refinement. We showcase xAI-CycleGAN's efficacy in maintaining structural integrity and generating high-quality images. Additionally, a histopathologist survey indicates the generated images' realism is often comparable to actual images, validating our model's high-quality output.
|
cs
|
2,312.04739 |
Natural gradient descent has a remarkable property that in the small learning rate limit, it displays an invariance with respect to network reparameterizations, leading to robust training behavior even for highly covariant network parameterizations. We show that optimization algorithms with this property can be viewed as discrete approximations of natural transformations from the functor determining an optimizer's state space from the diffeomorphism group if its configuration manifold, to the functor determining that state space's tangent bundle from this group. Algorithms with this property enjoy greater efficiency when used to train poorly parameterized networks, as the network evolution they generate is approximately invariant to network reparameterizations. More specifically, the flow generated by these algorithms in the limit as the learning rate vanishes is invariant under smooth reparameterizations, the respective flows of the parameters being determined by equivariant maps. By casting this property a natural transformation, we allow for generalizations beyond equivariance with respect to group actions; this framework can account for non-invertible maps such as projections, creating a framework for the direct comparison of training behavior across non-isomorphic network architectures, and the formal examination of limiting behavior as network size increases by considering inverse limits of these projections, should they exist. We introduce a simple method of introducing this naturality more generally and examine a number of popular machine learning training algorithms, finding that most are unnatural.
|
cs
|
2,312.03381 |
We report the discovery of a geocoronal solar wind charge exchange (SWCX) event corresponding to the well-known 2006 December 13th coronal mass ejection (CME) event. Strong evidence for the charge exchange origin of this transient diffuse emission is provided by prominent non-thermal emission lines at energies of $\rm O^{7+}$, $\rm Ne^{9+}$, $\rm Mg^{11+}$, $\rm Si^{12+}$, $\rm Si^{13+}$. Especially, a 0.53 keV emission line that most likely arises from the $\rm N^{5+}$ $1s^1 5p^1 \to 1s^2$ transition is detected. Previously, the forecastability of SWCX occurrence with proton flares has been disputed. In this particular event, we found that the SWCX signal coincided with the arrival of the magnetic cloud inside CME, triggered with a time delay after the proton flux fluctuation as the CME shock front passed through the Earth. Moreover, a spacecraft orbital modulation in SWCX light curve suggests that the emission arises close to the Earth. The line of sight was found to always pass through the northern magnetospheric cusp. The SWCX intensity was high when the line of sight passed the dusk side of the cusp, suggesting an azimuthal anisotropy in the flow of solar-wind ions inside the cusp. An axisymmetric SWCX emission model is found to underestimate the observed peak intensity by a factor of about 50. We suggest this discrepancy is related to the azimuthal anisotropy of the solar-wind flow in the cusp.
|
astro-ph
|
2,312.05662 |
Large Language Models (LLMs) trained with self-supervision on vast corpora of web text fit to the social biases of that text. Without intervention, these social biases persist in the model's predictions in downstream tasks, leading to representational harm. Many strategies have been proposed to mitigate the effects of inappropriate social biases learned during pretraining. Simultaneously, methods for model compression have become increasingly popular to reduce the computational burden of LLMs. Despite the popularity and need for both approaches, little work has been done to explore the interplay between these two. We perform a carefully controlled study of the impact of model compression via quantization and knowledge distillation on measures of social bias in LLMs. Longer pretraining and larger models led to higher social bias, and quantization showed a regularizer effect with its best trade-off around 20% of the original pretraining time.
|
cs
|
2,312.02867 |
The Health Index (HI) is crucial for evaluating system health, aiding tasks like anomaly detection and predicting remaining useful life for systems demanding high safety and reliability. Tight monitoring is crucial for achieving high precision at a lower cost, with applications such as spray coating. Obtaining HI labels in real-world applications is often cost-prohibitive, requiring continuous, precise health measurements. Therefore, it is more convenient to leverage run-to failure datasets that may provide potential indications of machine wear condition, making it necessary to apply semi-supervised tools for HI construction. In this study, we adapt the Deep Semi-supervised Anomaly Detection (DeepSAD) method for HI construction. We use the DeepSAD embedding as a condition indicators to address interpretability challenges and sensitivity to system-specific factors. Then, we introduce a diversity loss to enrich condition indicators. We employ an alternating projection algorithm with isotonic constraints to transform the DeepSAD embedding into a normalized HI with an increasing trend. Validation on the PHME 2010 milling dataset, a recognized benchmark with ground truth HIs demonstrates meaningful HIs estimations. Our methodology is then applied to monitor wear states of thermal spray coatings using high-frequency voltage. Our contributions create opportunities for more accessible and reliable HI estimation, particularly in cases where obtaining ground truth HI labels is unfeasible.
|
cs
|
2,312.05063 |
The growth and progression of brain tumors is governed by patient-specific dynamics. Even when the tumor appears well-delineated in medical imaging scans, tumor cells typically already have infiltrated the surrounding brain tissue beyond the visible lesion boundaries. Quantifying and understanding these growth dynamics promises to reveal this otherwise hidden spread and is key to individualized therapies. Current treatment plans for brain tumors, such as radiotherapy, typically involve delineating a standard uniform margin around the visible tumor on imaging scans to target this invisible tumor growth. This "one size fits all" approach is derived from population studies and often fails to account for the nuances of individual patient conditions. Here, we present the framework GliODIL which infers the full spatial distribution of tumor cell concentration from available imaging data based on PDE-constrained optimization. The framework builds on the newly introduced method of Optimizing the Discrete Loss (ODIL), data are assimilated in the solution of the Partial Differential Equations (PDEs) by optimizing a cost function that combines the discrete form of the equations and data as penalty terms. By utilizing consistent and stable discrete approximations of the PDEs, employing a multigrid method, and leveraging automatic differentiation, we achieve computation times suitable for clinical application such as radiotherapy planning. Our method performs parameter estimation in a manner that is consistent with the PDEs. Through a harmonious blend of physics-based constraints and data-driven approaches, GliODIL improves the accuracy of estimating tumor cell distribution and, clinically highly relevant, also predicting tumor recurrences, outperforming all other studied benchmarks.
|
cs
|
2,312.03498 |
The physics of primordial black holes can be affected by the non-Gaussian statistics of the density fluctuations that generate them. Therefore, it is important to have good theoretical control of the higher-order correlation functions for primordial curvature perturbations. By working at leading order in a $1/|\eta|$ expansion, we analytically determine the bispectrum of curvature fluctuations for single field inflationary scenarios producing primordial black holes. The bispectrum has a rich scale and shape dependence, and its features depend on the dynamics of the would-be decaying mode. We apply our analytical results to study gravitational waves induced at second order by enhanced curvature fluctuations. Their statistical properties are derived in terms of convolution integrals over wide momentum ranges, and they are sensitive on the scale and shape dependence of the curvature bispectrum we analytically computed.
|
astro-ph
|
2,312.04427 |
Recent molecular communication (MC) research has integrated more detailed computational models to capture the dynamics of practical biophysical systems. This research focuses on developing realistic models for MC transceivers inspired by spheroids - three-dimensional cell aggregates commonly used in organ-on-chip experimental systems. Potential applications that can be used or modeled with spheroids include nutrient transport in an organ-on-chip system, the release of biomarkers or reception of drug molecules by a cancerous tumor site, or transceiver nanomachines participating in information exchange. In this paper, a simple diffusive MC system is considered where a spheroidal transmitter and receiver are in an unbounded fluid environment. These spheroidal antennas are modeled as porous media for diffusive signaling molecules, then their boundary conditions and effective diffusion coefficients are characterized. Further, for either a point source or spheroidal transmitter, Green's function for concentration (GFC) outside and inside the receiving spheroid is analytically derived and formulated in terms of an infinite series and confirmed by a particle-based simulator (PBS). The provided GFCs enable computation of the transmitted and received signals in the spheroidal communication system. This study shows that the porous structure of the receiving spheroid amplifies diffusion signals but also disperses them, thus there is a trade-off between porosity and information transmission rate. Also, the results reveal that the porous arrangement of the transmitting spheroid not only disperses the received signal but also attenuates it. System performance is also evaluated in terms of bit error rate (BER). Decreasing the porosity of the receiving spheroid is shown to enhance system performance. Conversely, reducing the porosity of the transmitting spheroid can adversely affect system performance.
|
math
|
2,312.05253 |
We develop a generative attention-based approach to modeling structured entities comprising different property types, such as numerical, categorical, string, and composite. This approach handles such heterogeneous data through a mixed continuous-discrete diffusion process over the properties. Our flexible framework can model entities with arbitrary hierarchical properties, enabling applications to structured Knowledge Base (KB) entities and tabular data. Our approach obtains state-of-the-art performance on a majority of cases across 15 datasets. In addition, experiments with a device KB and a nuclear physics dataset demonstrate the model's ability to learn representations useful for entity completion in diverse settings. This has many downstream use cases, including modeling numerical properties with high accuracy - critical for science applications, which also benefit from the model's inherent probabilistic nature.
|
cs
|
2,312.05382 |
Differentiating noisy, discrete measurements in order to fit an ordinary differential equation can be unreasonably effective. Assuming square-integrable noise and minimal flow regularity, we construct and analyze a finite-difference differentiation filter and a Tikhonov-regularized least squares estimator for the continuous-time parameter-linear system. Combining these contributions in series, we obtain a finite-sample bound on mean absolute error of estimation. As a by-product, we offer a novel analysis of stochastically perturbed Moore-Penrose pseudoinverses.
|
math
|
2,312.03537 |
We study analysis on the cone of discrete Radon measures over a locally compact Polish space $X$. We discuss probability measures on the cone and the corresponding correlation measures and correlation functions on the sub-cone of finite discrete Radon measures over $X$. For this, we consider on the cone an analogue of the harmonic analysis on the configuration space developed in [12]. We also study elements of the difference calculus on the cone: we introduce discrete birth-and-death gradients and study the corresponding Dirichlet forms; finally, we discuss a system of polynomial functions on the cone which satisfy the binomial identity.
|
math
|
2,312.03986 |
With the ever-increasing demand for high-speed wireless data transmission, beamforming techniques have been proven to be crucial in improving the data rate and the signal-to-noise ratio (SNR) at the receiver. However, they require feedback mechanisms that need an overhead of information and increase the system complexity, potentially challenging the efficiency and capacity of modern wireless networks. This paper investigates novel index-based feedback mechanisms that aim at reducing the beamforming feedback overhead in Wi-Fi links. The proposed methods mitigate the overhead by generating a set of candidate beamforming vectors using an unsupervised learning-based framework. The amount of feedback information required is thus reduced by using the index of the candidate as feedback instead of transmitting the entire beamforming matrix. We explore several methods that consider different representations of the data in the candidate set. In particular, we propose five different ways to generate and represent the candidate sets that consider the covariance matrices of the channel, serialize the feedback matrix, and account for the effective distance, among others. Additionally, we also discuss the implications of using partial information in the compressed beamforming feedback on the link performance and compare it with the newly proposed index-based methods. Extensive IEEE 802.11 standard-compliant simulation results show that the proposed methods effectively minimize the feedback overhead, enhancing the throughput while maintaining an adequate link performance.
|
eess
|
2,312.0489 |
Distributionally robust optimization is used to solve decision making problems under adversarial uncertainty where the distribution of the uncertainty is itself ambiguous. In this paper, we identify a class of these instances that is solvable in polynomial time by viewing it through the lens of submodularity. We show that the sharpest upper bound on the expectation of the maximum of affine functions of a random vector is computable in polynomial time if each random variable is discrete with finite support and upper bounds (respectively lower bounds) on the expected values of a finite set of submodular (respectively supermodular) functions of the random vector are specified. This adds to known polynomial time solvable instances of the multimarginal optimal transport problem and the generalized moment problem by bridging ideas from convexity in continuous optimization to submodularity in discrete optimization. In turn, we show that a class of distributionally robust optimization problems with discrete random variables is solvable in polynomial time using the ellipsoid method. When the submodular (respectively supermodular) functions are structured, the sharp bound is computable by solving a compact linear program. We illustrate this in two cases. The first is a multimarginal optimal transport problem with given univariate marginal distributions and bivariate marginals satisfying specific positive dependence orders along with an extension to incorporate higher order marginal information. The second is a discrete moment problem where a set of marginal moments of the random variables are given along with lower bounds on the cross moments of pairs of random variables. Numerical experiments show that the bounds improve by 2 to 8 percent over bounds that use only univariate information in the first case, and by 8 to 15 percent over bounds that use the first moment in the second case.
|
math
|
2,312.05335 |
The coherence characteristics of a tin-vacancy color center in diamond are investigated through optical means including coherent population trapping between the ground state orbital levels and linewidth broadening effects. Due to the large spin-orbit splitting of the orbital ground states, thermalization between the ground states occurs at rates that are impractical to measure directly. Here, spectral information is transformed into its conjugate variable time, providing picosecond resolution and revealing an orbital depolarization timescale of ${\sim30{\rm~ps}}$. Consequences of the investigated dynamics are then used to estimate spin dephasing times limited by thermal effects.
|
quant-ph
|
2,312.04107 |
Quantum key distribution (QKD) protocols are essential to guarantee information-theoretic security in quantum communication. Although there was some previous work on quantum group key distribution, they still face many challenges under a ``\textit{dynamic}'' group communication scenario. In particular, when the group keys need to be updated in real-time for each user joining or leaving to ensure secure communication properties, i.e., forward confidentiality and backward confidentiality. However, current protocols require a large amount of quantum resources to update the group keys, and this makes them impractical for handling large and dynamic communication groups. In this paper, we apply the notion of ``{\em tree key graph}'' to the quantum key agreement and propose two dynamic Quantum Group Key Agreement (QGKA) protocols for a join or leave request in group communications. In addition, we analyze the quantum resource consumption of our proposed protocols. The number of qubits required per join or leave only increases logarithmically with the group size. As a result, our proposed protocols are more practical and scalable for large and dynamic quantum group communications.
|
quant-ph
|
2,312.05736 |
Graph Comparative Learning (GCL) is a self-supervised method that combines the advantages of Graph Convolutional Networks (GCNs) and comparative learning, making it promising for learning node representations. However, the GCN encoders used in these methods rely on the Fourier transform to learn fixed graph representations, which is inherently limited by the uncertainty principle involving spatial and spectral localization trade-offs. To overcome the inflexibility of existing methods and the computationally expensive eigen-decomposition and dense matrix multiplication, this paper proposes an Adaptive Spectral Wavelet Transform-based Self-Supervised Graph Neural Network (ASWT-SGNN). The proposed method employs spectral adaptive polynomials to approximate the filter function and optimize the wavelet using contrast loss. This design enables the creation of local filters in both spectral and spatial domains, allowing flexible aggregation of neighborhood information at various scales and facilitating controlled transformation between local and global information. Compared to existing methods, the proposed approach reduces computational complexity and addresses the limitation of graph convolutional neural networks, which are constrained by graph size and lack flexible control over the neighborhood aspect. Extensive experiments on eight benchmark datasets demonstrate that ASWT-SGNN accurately approximates the filter function in high-density spectral regions, avoiding costly eigen-decomposition. Furthermore, ASWT-SGNN achieves comparable performance to state-of-the-art models in node classification tasks.
|
eess
|
2,312.04658 |
Inductive Conformal Prediction (ICP) provides a practical and effective approach for equipping deep learning models with uncertainty estimates in the form of set-valued predictions which are guaranteed to contain the ground truth with high probability. Despite the appeal of this coverage guarantee, these sets may not be efficient: the size and contents of the prediction sets are not directly controlled, and instead depend on the underlying model and choice of score function. To remedy this, recent work has proposed learning model and score function parameters using data to directly optimize the efficiency of the ICP prediction sets. While appealing, the generalization theory for such an approach is lacking: direct optimization of empirical efficiency may yield prediction sets that are either no longer efficient on test data, or no longer obtain the required coverage on test data. In this work, we use PAC-Bayes theory to obtain generalization bounds on both the coverage and the efficiency of set-valued predictors which can be directly optimized to maximize efficiency while satisfying a desired test coverage. In contrast to prior work, our framework allows us to utilize the entire calibration dataset to learn the parameters of the model and score function, instead of requiring a separate hold-out set for obtaining test-time coverage guarantees. We leverage these theoretical results to provide a practical algorithm for using calibration data to simultaneously fine-tune the parameters of a model and score function while guaranteeing test-time coverage and efficiency of the resulting prediction sets. We evaluate the approach on regression and classification tasks, and outperform baselines calibrated using a Hoeffding bound-based PAC guarantee on ICP, especially in the low-data regime.
|
stat
|
2,312.03304 |
This letter establishes a novel relationship between a class of recurrent neural networks and certain evolutionary dynamics that emerge in the context of population games. Specifically, it is shown that the output of a recurrent neural network, in the context of classification problems, coincides with the evolution of the population state in a population game. This connection is established with dynamic payoffs and under replicator evolutionary dynamics. The connection provides insights into the neural network's behavior from both dynamical systems and game-theoretical perspectives, aligning with recent literature that suggests that neural network outputs may resemble the Nash equilibria of suitable games. It also uncovers potential connections between the neural network classification problem and mechanism design. To illustrate our results, we present different numerical experiments in the context of classification problems.
|
math
|
2,312.04903 |
The real network has two characteristics: heterogeneity and homogeneity. A directed network model with covariates is proposed to analyze these two features, and the asymptotic theory of parameter Maximum likelihood estimators(MLEs) is established. However, in many practical cases, network data often carries a lot of sensitive information. How to achieve the trade-off between privacy and utility has become an important issue in network data analysis. In this paper, we study a directed $\beta$-model with covariates under differential privacy mechanism. It includes $2n$-dimensional node degree parameters $\boldsymbol{\theta}$ and a $p$-dimensional homogeneity parameter $\boldsymbol{\gamma}$ that describes the covariate effect. We use the discrete Laplace mechanism to release noise for the bi-degree sequences. Based on moment equations, we estimate the parameters of both degree heterogeneity and homogeneity in the model, and derive the consistency and asymptotic normality of the differentially private estimators as the number of nodes tends to infinity. Numerical simulations and case studies are provided to demonstrate the validity of our theoretical results.
|
math
|
2,312.05014 |
It is known that each knot has a semimeander diagram (i. e. a diagram composed of two smooth simple arcs), however the number of crossings in such a diagram can only be roughly estimated. In the present paper we provide a new estimate of the complexity of the semimeander diagrams. We prove that for each knot $K$ with more than 10 crossings, there exists a semimeander diagram with no more than $0.31 \cdot 1.558^{\operatorname{cr}(K)}$ crossings, where $\operatorname{cr}(K)$ is the crossing number of $K$. As a corollary, we provide new estimates of the complexity of other meander-like types of knot diagrams, such as meander diagrams and potholders. We also describe an efficient algorithm for constructing a semimeander diagram from a given one.
|
math
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.