arxiv_id
string | latex
string | image
image |
|---|---|---|
1703.00564v3
|
\begin{table}[H]
\small
\centering
\begin{threeparttable}
\caption{Time Profile for Tox21, MUV, QM8 and Lipophilicity(second)}
\begin{tabular}{ |c|c|c|c|c| }
\hline
\textbf{Model} & \textbf{Tox21} & \textbf{MUV} & \textbf{QM8} & \textbf{Lipophilicity} \\
\hline
\hline
Logreg & 93 & 522 & &\\
\hline
KernelSVM & 2574 & 2231 & & \\
\hline
KRR & & & 3390/5153* & 24\\
\hline
RF & 24273 & & & 186\\
\hline
XGBoost & 2082 & 2418 & & 410\\
\hline
Multitask/Singletask & 22 & 858 & 275/701* & 21\\
\hline
Bypass & 31 & 938 & &\\
\hline
IRV & 58 & 2674 & &\\
\hline
GC & 246 & 2320 & 512 & 131\\
\hline
Weave & 323 & 4593 & & 255\\
\hline
DAG & & & & 5142\\
\hline
DTNN & & & 940 &\\
\hline
MPNN & & & 3383 & 1626\\
\hline
\end{tabular}
\label{tab:running_time}
\begin{tablenotes}
\item * ECFP/Coulomb Matrix
\end{tablenotes}
\end{threeparttable}
\end{table}
| |
1703.00564v3
|
\begin{table}[H]
\small
\centering
\caption{PDBbind Performances (Root-Mean-Square Error)}
\begin{tabular}{ |c|c|c|c|c| }
\hline
\textbf{Model} & \textbf{Model} & \textbf{Training} & \textbf{Validation} & \textbf{Test} \\
\hline
\hline
\multirow{5}{*}{PDBbind - core}
& RF & $0.82\pm0.00$ & $2.02\pm0.02$ & $2.03\pm0.01$ \\\cline{2-5}
& RF(grid) & $0.73\pm0.01$ & $1.98\pm0.01$ & $2.27\pm0.01$ \\\cline{2-5}
& Multitask & $1.62\pm0.03$ & $\mathbf{1.86\pm0.01}$ & $2.21\pm0.02$ \\\cline{2-5}
& Multitask(grid) & $1.51\pm0.05$ & $1.92\pm0.02$ & $2.20\pm0.03$ \\\cline{2-5}
& GC & $1.42\pm0.04$ & $2.10\pm0.05$ & $\mathbf{1.92\pm0.07}$ \\\cline{2-5}
\hline
\hline
\multirow{5}{*}{PDBbind - refined}
& RF & $0.66\pm0.00$ & $1.48\pm0.00$ & $1.62\pm0.00$ \\\cline{2-5}
& RF(grid) & $0.51\pm0.00$ & $\mathbf{1.37\pm0.00}$ & $\mathbf{1.38\pm0.00}$ \\\cline{2-5}
& Multitask & $1.09\pm0.01$ & $1.53\pm0.03$ & $1.66\pm0.05$ \\\cline{2-5}
& Multitask(grid) & $0.55\pm0.02$ & $1.41\pm0.02$ & $1.46\pm0.05$ \\\cline{2-5}
& GC & $1.20\pm0.01$ & $1.55\pm0.05$ & $1.65\pm0.03$ \\\cline{2-5}
\hline
\hline
\multirow{5}{*}{PDBbind - full}
& RF & $0.66\pm0.00$ & $1.40\pm0.00$ & $1.31\pm0.00$ \\\cline{2-5}
& RF(grid) & $0.51\pm0.00$ & $\mathbf{1.35\pm0.00}$ & $\mathbf{1.25\pm0.00}$ \\\cline{2-5}
& Multitask & $1.52\pm0.17$ & $1.42\pm0.05$ & $1.45\pm0.14$ \\\cline{2-5}
& Multitask(grid) & $0.39\pm0.01$ & $1.40\pm0.03$ & $1.28\pm0.02$ \\\cline{2-5}
& GC & $1.65\pm0.10$ & $1.57\pm0.20$ & $1.44\pm0.12$ \\\cline{2-5}
\hline
\end{tabular}
\label{tab:PDBbind}
\end{table}
| |
1703.00564v3
|
\begin{table}[H]
\small
\centering
\caption{QM7b Test Set Performances of All Tasks(Mean Absolute Error)}
\begin{tabular}{ |c|c|c|c| }
\hline
\textbf{Task} & Multitask(CM) & KRR(CM) & DTNN \\
\hline
\hline
Atomization energy - PBE0 & $36.0$ & $\mathbf{9.3}$ & $21.5$\\
\hline
Excitation energy of maximal optimal absorption - ZINDO & $1.31$ & $1.83$ & $\mathbf{1.26}$\\
\hline
Highest absorption - ZINDO & $0.086$ & $0.098$ & $\mathbf{0.074}$\\
\hline
HOMO - ZINDO & $0.293$ & $0.369$ & $\mathbf{0.192}$\\
\hline
LUMO - ZINDO & $0.255$ & $0.361$ & $\mathbf{0.159}$\\
\hline
1st excitation energy - ZINDO & $0.368$ & $0.479$ & $\mathbf{0.296}$\\
\hline
Ionization potential - ZINDO & $0.305$ & $0.408$ & $\mathbf{0.214}$\\
\hline
Electron Affinity - ZINDO & $0.271$ & $0.404$ & $\mathbf{0.174}$\\
\hline
HOMO - KS & $0.247$ & $0.272$ & $\mathbf{0.155}$\\
\hline
LUMO - KS & $0.187$ & $0.239$ & $\mathbf{0.129}$\\
\hline
HOMO - GW & $0.270$ & $0.294$ & $\mathbf{0.166}$\\
\hline
LUMO - GW & $0.172$ & $0.236$ & $\mathbf{0.139}$\\
\hline
Polarizability - PBE0 & $0.335$ & $0.225$ & $\mathbf{0.173}$\\
\hline
Polarizability - SCS & $0.317$ & $\mathbf{0.116}$ & $0.149$\\
\hline
\end{tabular}
\label{tab:QM7b}
\end{table}
| |
1703.00564v3
|
\begin{table}[H]
\small
\centering
\caption{QM8 Test Set Performances of All Tasks(Mean Absolute Error)}
\begin{tabular}{ |c|c|c|c|c|c|c|c| }
\hline
\textbf{Task} & Multitask & GC & KRR & Multitask(CM) & KRR(CM) & DTNN & MPNN\\
\hline
\hline
E1 - CC2 & $0.0088$ & $\mathbf{0.0074}$ & $0.0115$ & $0.0125$ & $0.0137$ & $0.0092$ & $0.0084$\\
\hline
E2 - CC2 & $0.0098$ & $\mathbf{0.0085}$ & $0.0116$ & $0.0114$ & $0.0124$ & $0.0092$ & $0.0091$\\
\hline
f1 - CC2 & $\mathbf{0.0145}$ & $0.0175$ & $0.0202$ & $0.0186$ & $0.0272$ & $0.0182$ & $0.0151$\\
\hline
f2 - CC2 & $0.0320$ & $0.0328$ & $0.0387$ & $0.0358$ & $0.0460$ & $0.0377$ & $\mathbf{0.0314}$\\
\hline
E1 - PBE0 & $0.0089$ & $\mathbf{0.0076}$ & $0.0118$ & $0.0126$ & $0.0140$ & $0.0090$ & $0.0083$\\
\hline
E2 - PBE0 & $0.0096$ & $\mathbf{0.0083}$ & $0.0117$ & $0.0114$ & $0.0122$ & $0.0086$ & $0.0086$\\
\hline
f1 - PBE0 & $\mathbf{0.0121}$ & $0.0125$ & $0.0189$ & $0.0152$ & $0.0258$ & $0.0155$ & $0.0123$\\
\hline
f2 - PBE0 & $0.0252$ & $0.0246$ & $0.0319$ & $0.0267$ & $0.0376$ & $0.0281$ & $\mathbf{0.0236}$\\
\hline
E1 - CAM & $0.0083$ & $\mathbf{0.0070}$ & $0.0111$ & $0.0119$ & $0.0132$ & $0.0086$ & $0.0079$\\
\hline
E2 - CAM & $0.0090$ & $\mathbf{0.0076}$ & $0.0109$ & $0.0106$ & $0.0115$ & $0.0082$ & $0.0082$\\
\hline
f1 - CAM & $0.0140$ & $0.0153$ & $0.0208$ & $0.0177$ & $0.0304$ & $0.0180$ & $\mathbf{0.0134}$\\
\hline
f2 - CAM & $0.0274$ & $0.0285$ & $0.0345$ & $0.0303$ & $0.0417$ & $0.0322$ & $\mathbf{0.0258}$\\
\hline
\end{tabular}
\label{tab:QM8}
\end{table}
| |
1703.00564v3
|
\begin{table}[H]
\small
\centering
\caption{QM9 Test Set Performances of All Tasks(Mean Absolute Error)}
\begin{tabular}{ |c|c|c|c|c|c| }
\hline
\textbf{Task} & Multitask & Multitask(CM) & GC & DTNN & MPNN\\
\hline
\hline
mu & $0.602$ & $0.519$ & $0.583$ & $\mathbf{0.244}$ & $0.358$\\
\hline
alpha & $3.10$ & $\mathbf{0.85}$ & $1.37$ & $0.95$ & $0.89$\\
\hline
HOMO & $0.00660$ & $0.00506$ & $0.00716$ & $\mathbf{0.00388}$ & $0.00541$\\
\hline
LUMO & $0.00854$ & $0.00645$ & $0.00921$ & $\mathbf{0.00513}$ & $0.00623$\\
\hline
gap & $0.0100$ & $0.0086$ & $0.0112$ & $\mathbf{0.0066}$ & $0.0082$\\
\hline
R2 & $125.7$ & $46.0$ & $35.9$ & $\mathbf{17.0}$ & $28.5$\\
\hline
ZPVE & $0.01109$ & $0.00207$ & $0.00299$ & $\mathbf{0.00172}$ & $0.00216$\\
\hline
U0 & $15.10$ & $2.27$ & $3.41$ & $2.43$ & $\mathbf{2.05}$\\
\hline
U & $15.10$ & $2.27$ & $3.41$ & $2.43$ & $\mathbf{2.00}$\\
\hline
H & $15.10$ & $2.27$ & $3.41$ & $2.43$ & $\mathbf{2.02}$\\
\hline
G & $15.10$ & $2.27$ & $3.41$ & $2.43$ & $\mathbf{2.02}$\\
\hline
Cv & $1.77$ & $0.39$ & $0.65$ & $\mathbf{0.27}$ & $0.42$\\
\hline
\end{tabular}
\label{tab:QM9}
\end{table}
| |
1703.00564v3
|
\begin{table}[H]
\centering
\small
\caption{DeepChem commands to load MoleculeNet datasets and models}
\begin{tabular}{ |c|l| }
\hline
\textbf{Dataset} & \textbf{Command}\\
\hline
QM7 & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_qm7\_from\_mat}\\
\hline
QM7b & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_qm7b\_from\_mat}\\
\hline
QM8 & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_qm8}\\
\hline
QM9 & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_qm9}\\
\hline
ESOL & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_delaney}\\
\hline
FreeSolv & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_sampl}\\
\hline
Lipophilicity & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_lipo}\\
\hline
PCBA & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_pcba}\\
\hline
MUV & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_muv}\\
\hline
HIV & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_hiv}\\
\hline
BACE & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_bace\_classification}\\
\hline
PDBbind & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_pdbbind\_grid}\\
\hline
BBBP & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_bbbp}\\
\hline
Tox21 & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_tox21}\\
\hline
ToxCast & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_toxcast}\\
\hline
SIDER & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_sider}\\
\hline
ClinTox & {\fontfamily{pcr}\selectfont deepchem.molnet.load\_clintox}\\
\hline
\hline
\textbf{Model} & \textbf{Command} \\
\hline
Logreg$^a$ & {\fontfamily{pcr}\selectfont sklearn.linear\_model.LogisticRegression} \\
\hline
KernelSVM$^a$ & {\fontfamily{pcr}\selectfont sklearn.svm.SVC} \\
\hline
KRR$^a$ & {\fontfamily{pcr}\selectfont sklearn.kernel\_ridge.KernelRidge} \\
\hline
\multirow{2}{*}{RF$^a$} & {\fontfamily{pcr}\selectfont sklearn.ensemble.RandomForestClassifier} \\
& {\fontfamily{pcr}\selectfont sklearn.ensemble.RandomForestRegressor} \\
\hline
XGBoost$^b$ & {\fontfamily{pcr}\selectfont deepchem.models.xgboost\_models.XGBoostModel} \\
\hline
\multirow{2}{*}{Multitask/Singletask} & {\fontfamily{pcr}\selectfont deepchem.models.MultitaskClassifier} \\
& {\fontfamily{pcr}\selectfont deepchem.models.MultitaskRegressor} \\
\hline
Bypass & {\fontfamily{pcr}\selectfont deepchem.models.RobustMultitaskClassifier} \\
\hline
IRV & {\fontfamily{pcr}\selectfont deepchem.models.TensorflowMultitaskIRVClassifier} \\
\hline
GC & {\fontfamily{pcr}\selectfont deepchem.models.GraphConvModel}\\
\hline
Weave & {\fontfamily{pcr}\selectfont deepchem.models.WeaveModel}\\
\hline
DAG & {\fontfamily{pcr}\selectfont deepchem.models.DAGModel}\\
\hline
DTNN & {\fontfamily{pcr}\selectfont deepchem.models.DTNNModel}\\
\hline
ANI-1 & {\fontfamily{pcr}\selectfont deepchem.models.ANIRegression}\\
\hline
MPNN & {\fontfamily{pcr}\selectfont deepchem.models.MPNNModel}\\
\hline
\end{tabular}
\begin{tablenotes}
\item {$^a$} These models are based on scikit-learn package.\cite{sklearn}
\item {$^b$} XGBoost is based on xgboost package.\cite{xgb}
\end{tablenotes}
\label{tab:deepchem_command}
\end{table}
| |
2008.08516v3
|
\begin{table}[htbp]
\centering
\caption{Overview of main AutoML methdologies shown in chronological order. Please note that $\alpha-$level methods are not included in this table as they refer to any methodology for fitting a model to a dataset (e.g., least-squares for linear regression). }
\label{tab:historical}
\footnotesize{
\begin{tabular}{|p{1cm} |p{1cm} | p{1.7cm} | p{0.8cm} | p{3.5cm} |p{3.5cm} |}\hline
\textbf{Year}& \textbf{Ref.}& \textbf{Method} & \textbf{Type} & \textbf{Description} & \textbf{Innovative aspects} \\\hline
2006&\cite{psms06,Escalante:2009:PSM:1577069.1577084}& PSMS& $\beta$& Vectorial representation of solutions, PSO used as optimizer, subsampling, CV&Formulation of the full model selection task\\\hline
2007&\cite{DBLP:conf/cec/GorissenTCD08,DBLP:journals/jmlr/GorissenDT09}&Heterogeneous surrogate evolution&$\beta$&Parallel co evolution of models, ensemble generation& Returned ensemble of solutions, large and heterogenous space of models\\\hline
2010&\cite{DBLP:conf/ijcnn/EscalanteMS10}&Ensemble PSMS&$\beta$&Enhanced PSMS with ensemble of solutions& Returned an ensemble of solutions as output\\\hline
2012&\cite{10.1145/2330784.2331014}&GPS: GA-PSO-FMS& $\beta$ & GAs were used to search for a model template, PSO was used for hyperparameter optimization&Separation of template search and hyperparameter optimization\\\hline
2013&\cite{ThoHutHooLey13-AutoWEKA}&Auto-WEKA&$\gamma$&SMBO with SMAC, approached the CASH problem&Definition of the combined algorithm selection and hyperparameter optimization problem\\\hline
2014&\cite{DBLP:journals/ijon/Rosales-PerezGCEG14}&Multi-objective surrogate-based FMS&$\gamma$&Multi objective (complexity/performance) evolutionary method, surrogates were used to approximate the fitness function&Among the first methods using a meta-learner for AutoML, multi-objective formulation\\\hline
2015&\cite{Feurer2019}& AutoSkLearn & $\gamma$ & SMBO, warm starting with a classifier, ensemble generation & AutoML definition, warm-starting with meta-learner, winner of AutoML challenge\\\hline
2016&\cite{DBLP:conf/icml/OlsonM16,tpot}&TPOT&$\beta$&Genetic programming / NSGA-II selwection, cross validation, data sampling& Models naturally codified as GP trees \\\hline
%\textcolor{red}{2016}& \cite{DBLP:conf/aistats/JamiesonT16,10.5555/3122009.3242042}& \textcolor{red}{Hyperband}& $\beta$& \textcolor{red}{Random search with successive halving}&\textcolor{red}{Introduction of successive halving}\\\hline
2017-2020&\cite{DBLP:journals/corr/ZophL16,autokeras,10.5555/3305890.3305981,elsken2018neural,Elsken2019}&Neural Architecture search &$\gamma$&Reinforcement Learning, Evolutionary Algorithms, SMBO for Neural Architecture search&Novel codifications for architectures, comparison of architectures, \emph{ad-hoc} NAS surrogates\\\hline
%\cite{Feurer2019}& AutoSklearn & SMBO / warmstarting with a classifier& Predictor to estimate performance of candidate models & Ensemble generation\\\hline
%\cite{Escalante:2009:PSM:1577069.1577084,DBLP:conf/ijcnn/EscalanteMS10}&PSMS - EPSMS& Particle swarm optimizer& NA&Cross validation, subsampling, ensemble generation\\\hline
%\cite{tpot}&TPOT&Genetic programming / NSGA-II&NA&Cross validation, data sampling\\\hline
%\cite{smac}&SMAC&SMBO&Random forest&\\\hline
%>OP&&&\\\hline
%\cite{DBLP:journals/jmlr/GorissenDT09}&SUMO&Evolutionary algorithms&Regressor of performance&Ensemble generation\\\hline
%&AutoWEKA&&&\\\hline
%&AutoKeras&&&\\\hline
%&SMAC&&&\\\hline
\end{tabular}}
\end{table}
| |
2108.13581v1
|
\begin{table}[h]
\centering
\resizebox{0.47\textwidth}{!}{
\begin{tabular}{c|c|c|c|c}
\textit{Dataset} & \textit{records} & \textit{features} & \textit{after VIF} & \textit{outcome} \\
\hline
Synthetic & $5,000$ & $1$ & $1$ & Y \\
Metropolitan & $1,677$ & $42$ & $6$ & Valence \\
Wine Quality & $6,497$ & $11$ & $5$ & Quality \\
NYC & $36,805$ & $7$ & $4$ & Sale Price\\
Stack Overflow & $372,321$ & $13$ & $5$ & Answer Length \\%Words \\
\hline
\end{tabular}}
\caption{Data sets and their characteristics. \label{tab:info}}
\end{table}
| |
2006.02986v1
|
\begin{table}[ht]
\centering
\begin{tabular}{|c|c|c|}
\hline
Hyperparameter & Q-Network & EQLM \\ \hline
$\alpha$ & 0.0065 & - \\
$\bar{\gamma}$ & - & 1.827e-5 \\
$\Tilde{N}$ & 29 & 25 \\
$\epsilon_i$ & 0.670 & 0.559 \\
$N_{\epsilon}$ & 400 & 360 \\
$\gamma$ & 0.99 & 0.93 \\
$k$ & 26 & 2 \\
$C$ & 70 & 48 \\ \hline
\end{tabular}
\vspace{5pt}
\caption{Hyperparameters used for each agent in the cart-pole task}
\label{tab:params_cartpole}
\end{table}
| |
2006.02986v1
|
\begin{table}[ht]
\centering
\begin{tabular}{|c|c|c|c|}
\hline
\multicolumn{2}{|c|}{Measure} & Q-Network & EQLM \\ \hline
\multirow{2}{*}{reward} & mean & 160.0 (147.5, 173.7) & 166.9 (160.7, 173.3) \\
& std & 47.0 (35.1, 62.2) & 23.1 (20.3, 26.7) \\ \hline
\multirow{2}{*}{auc ($*10^3$)} & mean & 84.1 (81.0 87.4) & 83.3 (80.4, 86.2) \\
& std & 11.7 (9.1, 14.7) & 10.6 (9.3, 12.4) \\ \hline
\end{tabular}
\vspace{5pt}
\caption{Performance of each algorithm in the cart-pole task}
\label{tab:results_cartpole}
\end{table}
| |
2308.13645v1
|
\begin{table}[!h]
\centering
\begin{tabular}{ |c|c|c|c|c|c| }
\hline
\multirow{2}{*}{Sampling strategy} & \multicolumn{5}{|c|}{Number CRPs in training set} \\
\cline{2-6}
&$200$&$350$ & $550$ & $750$ & $1000$\\
\hline
Random &$14.6$&$8.8$&$5.6$&$4.0$&$3.2$\\\hline
Active &$9.0$&$3.0$&$1.4$&$1.00$&$0.6$\\\hline
\end{tabular}
\caption{Active / fast learning, no noise, prediction error (\%) using Algorithm ``Active learning'' with $k=0$}
\label{table:us-nonoise}
\end{table}
| |
2308.13645v1
|
\begin{table}[!h]
\centering
\begin{tabular}{ |c|c|c|c|c|c| }
\hline
\multirow{2}{*}{Sampling strategy} & \multicolumn{5}{|c|}{Number CRPs in training set} \\
\cline{2-6}
&$350$&$550$ & $1250$ & $1750$ & $6000$\\
\hline
Random &$11.5$&$8.10$&$6.12$&$5.90$&$4.30$\\\hline
EQB &$9.35$&$5.90$&$4.75$&$4.33$&$4.20$\\\hline
Error rate-based &$7.28$&$5.52$&$4.07$&$3.43$&$2.70$\\
\hline
\end{tabular}
\caption{Active / fast learning, work by \cite{active-PUF}: $3.5\%$ noise, prediction error $(\%)$}
\label{table:others-noise}
\end{table}
| |
2308.13645v1
|
\begin{table}[!h]
\centering
\begin{tabular}{ |c|c|c|c|c|c| }
\hline
\multirow{2}{*}{Sampling strategy} & \multicolumn{5}{|c|}{Number CRPs in training set} \\
\cline{2-6}
&$200$&$350$ & $550$ & $750$ & $1000$\\
\hline
Random &$15.9$&$10.0$&$7.37$&$6.46$&$5.6$\\\hline
Active &$10.4$&$5.4$&$3.9$&$3.6$&$3.5$\\\hline
\end{tabular}
\caption{Active / fast learning, $3.5\%$ noise, prediction error (\%) using Algorithm ``Active learning'' with $k=0$}
\label{table:us-noise}
\end{table}
| |
2308.13645v1
|
\begin{table}[!h]
\centering
\begin{tabular}{ |c|c|c|c|c| }
\hline
\multirow{2}{*}{ML method} & \multicolumn{4}{|c|}{Number CRPs in training set} \\
\cline{2-5}
&$1000$&$3000$ & $5000$& $10000$\\
\hline
LR &$63$&$63$&$63$&$64$\\\hline
SVM &$67$&$68$&$68$&$68$\\\hline
Nnet &$67$&$68$&$68$&$68$\\
\hline
Adaboost &$66$&$67$&$67$&$67$\\
\hline
\end{tabular}
\caption{Adversarial / slow learning, no noise, accuracy (\%) using Algorithm ``Active learning'' with $k=5$, $|S|=40$.}
\label{table:adversarial}
\end{table}
| |
2312.09352v1
|
\begin{table}[h!]
\label{tbl:ablation_study}
\centering
\caption{Average accuracy of different methods on\datasetTwo dataset with task size 20}
\begin{tabular}{*{3}{|c}|}
\hline
\bfseries Method & \bfseries \datasetTwo \\ \hline
baseline & 44.54 \\ \hline
\pbesShort w/o KeepAugment & 63.56 \\ \hline
baseline w/ KeepAugment & 61.28 \\ \hline
\pbesShort w/ KeepAugment & 68.27 \\ \hline
\end{tabular}
\end{table}
| |
2312.09352v1
|
\begin{table}[ht]
% % \centering
% \caption{Influence of Exemplars Size for (a)\datasetOne dataset \& (b)\datasetTwo dataset}
% \begin{subtable}[t]{0.50 \textwidth}
% \centering
% \caption{}
% \begin{tabular}{ |c|c|c|c| }
% % \centering
% \hline
% $M$=550 & $M$=700 & $M$=850 & $M$=1000 \\ \hline
% 68.27 & 71.43 & 75.1 & 78.3 \\ \hline
% \end{tabular}
% \label{tbl:exemplars_effect_tbl1}%
% \end{subtable}
% \begin{subtable}[t]{0.50 \textwidth}
% \centering
% \caption{}
% \begin{tabular}{ |c|c|c|c| }
% \hline
% $M$=800 & $M$=1000 & $M$=1200 & $M$=1400 \\ \hline
% 63.97 & 66.4 & 69.2 & 72.23 \\ \hline
% \end{tabular}
% \label{tbl:exemplars_effect_tbl2}%
% \end{subtable}
% \end{table}
| |
2209.07657v1
|
\begin{table}[!ht]
\centering
\caption{Slopes for Small and Large Saccades}
\label{tab:slopes}
\begin{tabular}{|c|c|c|c|c|}
\hline
Filter & Mean & SD & Mean & SD \\
& Small & Small & Large & Large \\ \hline
No Filter & 0.673 & 0.089 & 0.438 & 0.117 \\ \hline
STD & 0.670 & 0.086 & 0.445 & 0.115 \\ \hline
EXTRA & 0.638 & 0.084 & 0.445 & 0.111 \\ \hline
Z-LP100 & 0.686 & 0.101 & 0.439 & 0.122 \\ \hline
Z-LP50 & 0.780 & 0.092 & 0.456 & 0.108 \\ \hline
\end{tabular}
\end{table}
| |
2209.07657v1
|
\begin{table}[!ht]
\centering
\caption{Saccade Amplitudes and Peak Velocities for Very Large Saccades}
\label{tab:HOR}
\begin{tabular}{|c|c|c|c|c|c|}
\hline
FiltGrp & Count & Mean & SD & Mean & SD \\
& & Ampl & Ampl & PkVel & PkVel \\ \hline
No Filt & 1059 & 28.26 & 2.09 & 531.15 & 77.01 \\ \hline
STD & 1069 & 28.26 & 2.10 & 530.82 & 77.50 \\ \hline
EXTRA & 1084 & 28.24 & 2.09 & 530.46 & 77.50 \\ \hline
Z-LP100 & 1084 & 28.27 & 2.10 & 523.21 & 78.14 \\ \hline
Z-LP50 & 1049 & 28.28 & 2.11 & 515.08 & 77.05 \\ \hline
\end{tabular}
\end{table}
| |
2002.10269v1
|
\begin{table}[h]
\centering
\footnotesize
\begin{tabular}{|c|c|}
\hline
Quantities& number \\
\hline
$|Q|$ number of sequences $\equiv$ drives &224265\\
\hline
$|C|$ number of distinct cycles &3767\\
\hline
$|P|$ number of simple paths &7438\\
\hline
$c_e = c_v$ total number of edges or vertices in cycles &16077\\
\hline
$p_e = p_v -1$ total number of edges or vertices in paths &25736\\
\hline
$V$ number of vertices in FSA $\equiv$ states &124\\
\hline
$E$ number of edges in FSA $\equiv$ transits &1691\\
\hline
\end{tabular}
\caption{Overview of the statistics for the prototyped subset.}
\label{tab:problem}
\end{table}
| |
2002.10269v1
|
\begin{table}[h]
\centering
\footnotesize
\begin{tabular}{|c|c|c|c|c|c|c|c|c|}
\hline
\multicolumn{5}{|c}{}&
\multicolumn{2}{|c|}{Seq. Components}&
\multicolumn{2}{c|}{}\\
\hline
&Vehicle &Drove &Drive &Has &Circles &Paths &State &Transit\\
\hline
NodesV2 &85975 &- &224265 &- &- &- &49375&-\\
\hline
EdgesV2 &- &224265 &- &1570774 &16077 &25736 &-&-\\
\hline
NodesV3 &85975 &- &224265 &- &3767 &7438 &-&1691\\
\hline
EdgesV3 &- & 224265 &- &1570774 &16077 &25736 &-&-\\
\hline
\end{tabular}
\caption{Model Variant 2 compared with Model Variant 3}
\label{tab:models}
\end{table}
| |
2005.01138v1
|
\begin{table}[h!]
\centering
\begin{tabular}{|l|c|r|}
\hline
Environment & Avg returns over seed 0-9 & Temperature parameter\\
\hline
HalfCheetah-v2 & 11083 & fixed 0.2 \\
\hline
Ant-v2 & 4294 & learned \\
\hline
Hopper-v2 & 3544 & fixed 0.2 \\
\hline
Walker2d-v2 & 4490 & fixed 0.2\\
\hline
CustomAnt-v0 & 2430 & fixed 0.2 \\
\hline
DisabledAnt-v0 & 1089 & learned \\
\hline
PointMazeLeft-v0 & -7.37 & learned \\
\hline
PointMazeRight-v0 & -7.33 & learned \\
\hline
\end{tabular}
\caption{Expert Average Performance over 0-9 seed after $1e^6$ iterations.}
\label{expert_performance}
\end{table}
| |
0909.3691v1
|
\begin{table}[!t]
\caption{Genotype-Phenotype Connections in Genetic Disorders}
\centering
\begin{tabular}{ | c || p{2.2cm} || p{1.4cm} || p{2.2cm} | }%
\hline
\bf{Alleles} & \bf{Genotype} & \bf{Dominant Disorder} & \bf{Recessive Disorder} \\[1ex]
\hline
&&&\\
\bf{AA} & Homozygous WT & Normal & Normal \\[1ex]
\bf{Aa} & Heterozygous & Affected & Carrier \\[1ex]
\bf{aa} & Homozygous mut. & Affected & Affected \\[1ex]
\hline
\multicolumn{3}{l} {A - normal allele, a - mutant allele}
\end{tabular}
\label{table_genotype_phentype}
\end{table}
| |
2202.07254v1
|
\begin{table}[thb]
\vspace{.1in}
\caption{Relative interaction importance on a node level (left) and on a feature level (right). Gray shadings indicate how $intImp_j$ is calculated from $intImp(\mathcal{N}_P)$. The parameters $d$ and $P$ indicate the tree depth and the index of the parent node, respectively.}
\vspace{.1in}
\label{tab:intImportanceExample}
\begin{center}
\begin{tabular}{cc}
\begin{tabular}{|c|c|c|c|}
\hline
d & P & $\xv_j$ & $intImp(\mathcal{N}_P)$ \\\hline
\rowcolor[gray]{.9}
0 & 1 & $\xv_3$ & 0.835 \\
\rowcolor[gray]{.6}
1 & 2 & $\xv_1$ & 0.074\\
1 & 3 & $\xv_1$ & 0.066\\\hline
\end{tabular}\hspace*{0.5cm}
\begin{tabular}{|c|c|}
\hline
$\xv_j$ & $intImp_j$ \\\hline
\rowcolor[gray]{.9}
$\xv_3$ & 0.835 \\
\rowcolor[gray]{.6}
$\xv_1$ & 0.14\\\hline
\end{tabular}
\end{tabular}
\end{center}
\end{table}
| |
2202.07254v1
|
\begin{table}[thb]
\vspace{.1in}
\caption{Summary table of settings and key results of the simulation study in Section \ref{sec:sim_weak}. The column ``Setting'' refers to the setting number in Section \ref{sec:sim_weak}. The second column refers to the adjustments made in the setting compared to the initial setting. The other four columns show if the average ranks (r) of the feature interactions with the feature of interest ($\xv_2$) are correct (meaning that the ranks are the same as the ranks of the underlying data-generating process and fitted linear model) or if they are wrong (different from the ranks in the data-generating process and fitted linear model).}
\vspace{.1in}
\label{tab:simSummary}
\begin{center}
\begin{tabular}{|p{1cm}|p{4cm}|p{2.2cm}|p{2.2cm}|p{2.2cm}|p{2.2cm}|}
\hline
Setting & Adjustment & REPID & H-Statistic & Greenwell & Shapley \\\hline
(2) & $\beta_1 = 0.1$ (initial: 1) & \textbf{correct} $r(x_1) = r(x_3)$ & \textbf{wrong} $r(x_1) > r(x_3)$ & \textbf{wrong} $r(x_1) < r(x_3)$ & \textbf{correct}\\
(2) & $\beta_1 = \beta_2 = \beta_3 = \beta_4 = 0.1$ and $\beta_{12} = 2$ (initial: 1) & \textbf{correct} $r(x_1) > r(x_3)$ & \textbf{wrong} $r(x_1) = r(x_3)$ & \textbf{correct} & \textbf{correct}\\
(3) & $\rho_{12}$ = 0.9 (initial: 0) & \textbf{correct} $r(x_1) = r(x_3)$ & \textbf{wrong} $r(x_1) < r(x_3)$ & \textbf{correct} & \textbf{wrong} $r(x_1) > r(x_3)$\\
\hline
\end{tabular}
\end{center}
\end{table}
| |
1709.06990v1
|
\begin{table}
\protect\caption{The lower (LCB) and upper (UCB) compression bound constrain the algorithm by ensuring that the compression rate for each compressor is between the two values respectively. The table presents the LCB and UCB values (\%) along with the minimum and maximum number of rules used in the experiments. These set compression rates were enforced and the performance of \textit{PARSEC} was measured on several datasets. The values for the minimum and maximum number of rules were determined by additional trial runs. It was observed that a greater number of rules were needed to achieve higher compression rates.} \label{table:ratesAndRules}
\begin{centering}
\begin{tabular}{|c|c|p{1.6cm}|p{1.6cm}|}
\hline
\textbf{LCB} & \textbf{UCB} & \textbf{$compression
\allowbreak
Rules_{min}$} & \textbf{$compression
\allowbreak
Rules_{max}$}\tabularnewline
\hline
10 & 13 & 5 & 50\tabularnewline
15 & 18 & 10 & 70\tabularnewline
20 & 13 & 20 & 90\tabularnewline
25 & 28 & 40 & 120\tabularnewline
30 & 33 & 90 & 150\tabularnewline
50 & 53 & 350 & 500\tabularnewline
\hline
\end{tabular}
\par\end{centering}
\end{table}
| |
1709.06990v1
|
\begin{table}[!h]
\protect\caption{Average change in test accuracy (\%) across all the data sets when
PARSEC models were created for 75\% compression rate. } \label{table:75Rate}
\begin{centering}
\begin{tabular}{|l|c|c|c|}
\hline
\textbf{Algorithm} & \vtop{\hbox{\strut \textbf{Change in}}\hbox{\strut \textbf{Test Accuracy}}} & \vtop{\hbox{\strut \textbf{Original}}\hbox{\strut \textbf{Accuracy}}}& \vtop{\hbox{\strut \textbf{Compressed}}\hbox{\strut \textbf{Accuracy}}} \tabularnewline
\hline
LingPipe & -3.2 & 79.7 & 76.4\tabularnewline
MeaningCloud & -4.2 & 62.5 & 58.2\tabularnewline
SentiStrength & -6.9 & 67.8 & 60.8\tabularnewline
uClassify & -6.4 & 80.7 & 74.3\tabularnewline
\hline
\end{tabular}
\par\end{centering}
\end{table}
| |
1709.06990v1
|
\begin{table}[!h]
\begin{centering}
\begin{tabular}{|>{\centering}p{1.2cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|c|}
\hline
& \textbf{IN} & \textbf{LP} & \textbf{MC} & \textbf{S140} & \textbf{SS} & \textbf{ST} & \textbf{uC} & \textbf{V}\\
\hline
{AIV} & -0.9 & -0.0 & 0.4 & -2.1 & -0.3 & -3.0 & -1.9 & -0.9\tabularnewline\hline
{MI} & 0.1 & 0.3 & -0.7 & -1.9 & -0.4 & 0.5 & -0.4 & -1.4\tabularnewline\hline
{Digital Music} & -1.0 & 0.2 & -0.3 & -2.5 & -0.1 & -1.9 & -0.4 & -2.0\tabularnewline\hline
{Baby} & 0.0 & 0.4 & -0.2 & -2.7 & 0.0 & -1.2 & 0.3 & 0.2\tabularnewline\hline
{Patio \& Garden} & -0.1 & 0.4 & -0.2 & -2.1 & -0.2 & -0.7 & 0.9 & -0.9\tabularnewline\hline
{Automotive} & -1.1 & 0.4 & -0.5 & -2.1 & 0.0 & -1.1 & 0.7 & -0.9\tabularnewline\hline
{Pet Supplies} & -1.2 & 0.3 & -0.4 & -2.9 & -0.4 & 0.3 & 0.0 & -1.6\tabularnewline\hline
{Apps for Android} & -1.4 & 0.0 & -0.5 & -2.5 & -0.3 & -3.5 & 0.0 & -1.6\tabularnewline\hline
{Beauty} & -0.5 & 0.0 & 0.2 & -1.2 & 0.0 & -0.5 & -0.7 & -0.4\tabularnewline\hline
{Tools \& Home} & 0.8 & -0.1 & -1.0 & -2.3 & -0.6 & -1.7 & 0.5 & 0.0\tabularnewline\hline
{Toys \& Games} & -0.5 & 0.0 & -1.5 & -1.8 & 0.2 & -3.1 & -0.1 & -1.4\tabularnewline\hline
{Health \& Personal} & 0.4 & 0.1 & -0.4 & -1.6 & -0.3 & -1.4 & 0.5 & -0.4\tabularnewline\hline
\textbf{Average} & -0.4 & 0.1 & -0.4 & -2.1 & -0.2 & -1.4 & 0.0 & -0.9 \tabularnewline
\hline
\end{tabular}
\par\end{centering}
\protect\caption{Difference between the compressed and original test accuracy when a compression rate between 10\% and 13\% was imposed.} \label{table:compression10}
\end{table}
| |
1709.06990v1
|
\begin{table} [!h]
\begin{centering}
\begin{tabular}{|>{\centering}p{1.2cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|>{\centering}p{0.59cm}|c|}
\hline
& \textbf{IN} & \textbf{LP} & \textbf{MC} & \textbf{S140} & \textbf{SS} & \textbf{ST} & \textbf{uC} & \textbf{V}\\
\hline
{AIV} & -1.1 & -1.1 & -0.2 & -2.7 & 0.1 & -3.2 & -2.0 & -0.6\tabularnewline\hline
{MI} & 1.1 & 0.5 & -0.8 & -2.9 & -0.8 & 0.0 & -1.3 & -0.8\tabularnewline\hline
{Digital Music} & -1.6 & 0.4 & -0.7 & -3.6 & -0.5 & -2.3 & -1.2 & -2.1\tabularnewline\hline
{Baby} & 0.0 & -0.9 & -0.5 & -2.4 & 0.2 & -2.0 & 0.6 & -0.1\tabularnewline\hline
{Patio \& Garden} & -0.3 & -0.4 & -0.3 & -3.5 & -0.6 & -1.2 & 0.5 & -1.0\tabularnewline\hline
{Automotive} & -1.4 & 0.6 & -1.0 & -2.8 & -0.3 & -2.2 & 0.9 & -0.8\tabularnewline\hline
{Pet Supplies} & -1.1 & 0.6 & -0.2 & -3.8 & -0.6 & -1.3 & 0.2 & -2.9\tabularnewline\hline
{Apps for Android} & -1.7 & -0.3 & -0.9 & -4.2 & -0.5 & -3.4 & -1.0 & -3.2\tabularnewline\hline
{Beauty} & -0.9 & -0.3 & -0.7 & -3.5 & -0.2 & -1.3 & -1.8 & -1.2\tabularnewline\hline
{Tools \& Home} & -0.2 & 0.0 & 0.0 & -3.0 & -0.3 & -1.7 & 1.2 & 0.7\tabularnewline\hline
{Toys \& Games} & -1.3 & -0.2 & -1.2 & -1.6 & -0.1 & -1.1 & -0.8 & -1.8\tabularnewline\hline
{Health \& Personal} & 0.1 & 0.0 & -0.6 & -2.8 & -0.5 & -2.6 & 0.2 & -0.4\tabularnewline\hline
\textbf{Average} & -0.7 & 0.0 & -0.6 & -3.0 & -0.3 & -1.8 & -0.3 & -1.2 \tabularnewline
\hline
\end{tabular}
\par\end{centering}
\protect\caption{Difference between the compressed and original test accuracy when a compression rate between 15\% and 18\% was imposed. } \label{table:compression15}
\end{table}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.