Filters








1,679 Hits in 3.8 sec

CEB Improves Model Robustness [article]

Ian Fischer, Alexander A. Alemi
2020 arXiv   pre-print
We demonstrate that the Conditional Entropy Bottleneck (CEB) can improve model robustness. CEB is an easy strategy to implement and works in tandem with data augmentation procedures.  ...  We report results of a large scale adversarial robustness study on CIFAR-10, as well as the ImageNet-C Common Corruptions Benchmark, ImageNet-A, and PGD attacks.  ...  We demonstrate: • CEB models are easy to implement and train. • CEB models show improved generalization performance over deterministic baselines on CIFAR10 and ImageNet. • CEB models show improved robustness  ... 
arXiv:2002.05380v1 fatcat:5lpcw7udyffprjt5q2kev6brku

CEB Improves Model Robustness

Ian Fischer, Alexander A. Alemi
2020 Entropy  
We demonstrate that the Conditional Entropy Bottleneck (CEB) can not only scale up to large scale image classification tasks, but can additionally improve model robustness.  ...  CEB is an easy strategy to implement and works in tandem with data augmentation procedures.  ...  We demonstrate: • CEB models are easy to implement and train. • CEB models show improved generalization performance over deterministic baselines on CIFAR10 and ImageNet. • CEB models show improved robustness  ... 
doi:10.3390/e22101081 pmid:33286850 pmcid:PMC7597163 fatcat:v4tortdudbeu7db6mkz67pwzwq

The Conditional Entropy Bottleneck [article]

Ian Fischer
2020 arXiv   pre-print
We find strong empirical evidence supporting our hypothesis that MNI models improve on these problems of robust generalization.  ...  We experimentally test our hypothesis by comparing the performance of CEB models with deterministic models and Variational Information Bottleneck (VIB) models on a variety of different datasets and robustness  ...  We have shown empirically that simply by switching to CEB, models may substantially improve their robust generalization, including (RG1) higher accuracy, (RG2) better adversarial robustness, and (RG3)  ... 
arXiv:2002.05379v1 fatcat:ptki62qee5hg7j4qrvp7dzzaeu

The Conditional Entropy Bottleneck

Ian Fischer
2020 Entropy  
We find strong empirical evidence supporting our hypothesis that MNI models improve on these problems of robust generalization.  ...  We experimentally test our hypothesis by comparing the performance of CEB models with deterministic models and Variational Information Bottleneck (VIB) models on a variety of different datasets and robustness  ...  We have shown empirically that simply by switching to CEB, models may substantially improve their robust generalization, including (RG1) higher accuracy, (RG2) better adversarial robustness, and (RG3)  ... 
doi:10.3390/e22090999 pmid:33286768 fatcat:wc2qcb3ixja7xkan6kpotobd2q

A Closer Look at the Adversarial Robustness of Information Bottleneck Models [article]

Iryna Korshunova, David Stutz, Alexander A. Alemi, Olivia Wiles, Sven Gowal
2021 arXiv   pre-print
Previous works showed that the robustness of models trained with information bottlenecks can improve upon adversarial training.  ...  We study the adversarial robustness of information bottleneck models for classification.  ...  Acknowledgements We would like to thank Taylan Cemgil, Lucas Theis, Hubert Soyer, Jonas Degrave, and the wonderful people from the robustness teams at DeepMind for their help with this project, interesting  ... 
arXiv:2107.05712v1 fatcat:t5bunyiaazbxvdmf7wderuoqta

Compressive Visual Representations [article]

Kuang-Huei Lee, Anurag Arnab, Sergio Guadarrama, John Canny, Ian Fischer
2021 arXiv   pre-print
Our experiments confirm that adding compression to SimCLR and BYOL significantly improves linear evaluation accuracies and model robustness across a wide range of domain shifts.  ...  As Lipschitz continuity is closely related to robustness, this provides a new explanation for why compressed models are more robust.  ...  This provides an alternative explanation, in addition to the information-theoretic view [26, 27, 1, 2], for why CEB compression improves SimCLR model robustness.  ... 
arXiv:2109.12909v3 fatcat:cb3ngpw4bbegro3arubpsiqv4q

Improving Middle School Quality in Poor Countries

Patrick J. McEwan, Erin Murphy-Graham, David Torres Irribarra, Claudia Aguilar, Renán Rápalo
2015 Educational Evaluation and Policy Analysis  
This article evaluates the impact and cost-effectiveness of offering an innovative middle school model-the Sistema de Aprendizaje Tutorial (SAT)-to Honduran villages instead of traditional middle schools  ...  The article is one of the few studies to rigorously evaluate a scaled-up instructional reform in a poor country, implemented with an alternative model of teacher recruitment and contracting.  ...  Two Models for Rural Middle Schools The SAT The Honduran SAT model combines features of interventions discussed in the last section, including improved instructional materials and teacher training; flexible  ... 
doi:10.3102/0162373714527786 fatcat:5dpn25sqkrfyno5rg3zeb3weie

A channel-estimate-based decision feedback equalizer robust under impulsive noise

Konstantinos Pelekanakis, Mandar Chitre
2014 2014 Underwater Communications and Networking (UComms)  
A channel-estimate-based decision feedback equalizer (CEB DFE) robust under impulsive noise is presented for single-input multiple-output (SIMO) underwater acoustic communications.  ...  Channel estimation is performed via the improvedproportionate M-estimate affine projection algorithm (IPMAPA), a linear complexity algorithm which is robust against impulsive interference and exploits  ...  CONCLUSION An improved CEB DFE receiver robust in impulsive noise was proposed.  ... 
doi:10.1109/ucomms.2014.7017119 fatcat:722zv5ai3jeirpbeizgwfivq4y

Genetic links between post-reproductive lifespan and family size in Framingham

Xiaofei Wang, Sean G. Byars, Stephen C. Stearns
2013 Evolution, Medicine and Public Health  
That result was not robust to changes in the sample. Further studies on larger samples are needed to validate the antagonistic pleiotropy of these genes.  ...  The genetic correlation was large, highly significant and strongly negative in women (r G = À0.877, P = 0.009) in a model without covariates, but not in men (P = 0.777).  ...  When a nonlinear term for CEB was included, it significantly improved the model fit and became more significant than the linear term.  ... 
doi:10.1093/emph/eot013 pmid:24481203 pmcid:PMC3868361 fatcat:xj6wkhygdfhk3ngqvd3myhgjgu

CEBS update: curated toxicology database with enhanced tools for data integration

Cari Martini, Ying Frances Liu, Hui Gong, Nicole Sayers, German Segura, Jennifer Fostel
2021 Nucleic Acids Research  
The data are made available through the CEBS Homepage (https://cebs.niehs.nih.gov/cebs/), guided search applications, flat files on FTP (file transfer protocol), and APIs (application programming interface  ...  The Chemical Effects in Biological Systems database (CEBS) contains extensive toxicology study results and metadata from the Division of the National Toxicology Program (NTP) and other studies of environmental  ...  ACKNOWLEDGEMENTS We thank NTP for data, direction and support, and the CEBS curation and development teams who are stewards for the data, supported the use of the tools and created novel applications to  ... 
doi:10.1093/nar/gkab981 pmid:34751388 pmcid:PMC8728139 fatcat:3ll4hzz2s5hwdjvxtahvgasrsm

Comparison of HRV indices obtained from ECG and SCG signals from CEBS database

Szymon Siecinski, Ewaryst J. Tkacz, Pawel S. Kostka
2019 BioMedical Engineering OnLine  
Using robust model improved achieved correlation between HRV indices obtained from ECG and SCG signals except the R 2 values of pNN50 values in signals p001-p020 and for all analyzed signals.  ...  Correlation between HRV indices was calculated as coefficient of determination ( R 2 ) to determine goodness of fit to linear model.  ...  Using robust model fitting improves the correlation between HRV indices obtained from ECG and SCG except for p series and all analyzed signals due to the large number of outliers found in linear model.  ... 
doi:10.1186/s12938-019-0687-5 fatcat:jamkpy25s5gobefjoxik4mkply

Predictive Information Accelerates Learning in RL [article]

Kuang-Huei Lee, Ian Fischer, Anthony Liu, Yijie Guo, Honglak Lee, John Canny, Sergio Guadarrama
2020 arXiv   pre-print
We hypothesize that capturing the predictive information is useful in RL, since the ability to model what will happen next is necessary for success on many tasks.  ...  We show that PI-SAC agents can substantially improve sample efficiency over challenging baselines on tasks from the DM Control suite of continuous control environments.  ...  We show that future prediction is empirically superior in Section 4.2 and Section D. 39, 28] have demonstrated that image augmentation improves sample-efficiency and robustness of model-free continuous  ... 
arXiv:2007.12401v2 fatcat:kdji7srhmbbodolier3tbnb3ku

An empirical examination of customer advocacy influenced by engagement behaviour and predispositions of FinTech customers in India

Archana Nayak Kini, Savitha Basri
2022 F1000Research  
This is one of the first research studies to demonstrate and empirically validate the CEB model for the FinTech industry during the COVID-19 pandemic.  ...  Structural equation modelling (SEM) using smart PLS (partial least squares) 3.0 was applied to test the theoretical model.  ...  Theoretical model.  ... 
doi:10.12688/f1000research.74928.1 fatcat:eggvh6tb2fg5vikqmatknjnhea

Effects of customer engagement behavior

Katarzyna Żyminkowska, Tomasz Żyminkowski, Rafał Błaszczyk
2017 Journal of Economics and Management  
Originality/value/contribution -Our model of firm-level performance effects of CEB extends existing proposals and contributes to the knowledge on effective CEB management process in network organizations  ...  Findings -We propose the logically arranged CEB effects in the conceptual model integrated with marketing metrics which are related to the recent advances in customer equity and customer asset management  ...  We discuss main assumptions of our extended CEB consequences model in the following section.  ... 
doi:10.22367/jem.2017.28.08 fatcat:cnfrpzqxzrh6fo4uyu7ewqa5dy

Identifying Areas with Elevated Disease Incidence Rates Using Empirical Bayes Estimators

Owen J. Devine, Thomas A. Louis, M. Elizabeth Halloran
2010 Geographical Analysis  
Substantial improve- ment in estimation, as measured by average MSE, was evident even seat the EB and CEB estimators were based on incorrect distributional assump- tions.  ...  Elizabeth Halloran / 197 prior different from that on which the estimators were based, they should not be interpreted as a full analysis of the robustness of the EB and CEB methods to incorrect distributional  ... 
doi:10.1111/j.1538-4632.1996.tb00930.x fatcat:xehq4vpx3netjhusozjcxcmrdi
« Previous Showing results 1 — 15 out of 1,679 results