2024
Three-Dimensional Reconstruction Pre-Training as a Prior to Improve Robustness to Adversarial Attacks and Spurious Correlation
Yamada Y, Zhang F, Kluger Y, Yildirim I. Three-Dimensional Reconstruction Pre-Training as a Prior to Improve Robustness to Adversarial Attacks and Spurious Correlation. Entropy 2024, 26: 258. PMID: 38539769, PMCID: PMC10968904, DOI: 10.3390/e26030258.Peer-Reviewed Original ResearchAdversarial trainingPre-trainingAdversarial attacksAdversarial robustnessRobustness of image classifiersModel of human visionComputational model of human visionAdversarial examplesImage classifierWeight initializationDataset settingData augmentationBackground textureSpurious correlationsHuman visionModels of visionImprove robustnessDatasetRobustnessImage formationAttacksComputational modelTrainingShapeNetAdversary
2022
Deep unsupervised feature selection by discarding nuisance and correlated features
Shaham U, Lindenbaum O, Svirsky J, Kluger Y. Deep unsupervised feature selection by discarding nuisance and correlated features. Neural Networks 2022, 152: 34-43. PMID: 35500458, PMCID: PMC9526895, DOI: 10.1016/j.neunet.2022.04.002.Peer-Reviewed Original ResearchMeSH KeywordsCluster AnalysisConceptsUnsupervised feature selectionCorrelated featuresNuisance featuresFeature selectionReal-world datasetsAutoencoder architectureComplete featuresClustering resultsGraph LaplacianArchitectural designModern datasetsOptimization processDatasetDifferentiable approachUnderlying structureLarge numberSimilar approachFeaturesArchitectureLarge subsetSelectionCode