Matches in SemOpenAlex for { <https://semopenalex.org/work/W3175596306> ?p ?o ?g. }
- W3175596306 abstract "Robust training methods against perturbations to the input data have received great attention in the machine learning literature. A standard approach in this direction is adversarial training which learns a model using adversarially-perturbed training samples. However, adversarial training performs suboptimally against perturbations structured across samples such as universal and group-sparse shifts that are commonly present in biological data such as gene expression levels of different tissues. In this work, we seek to close this optimality gap and introduce Group-Structured Adversarial Training (GSAT) which learns a model robust to perturbations structured across samples. We formulate GSAT as a non-convex concave minimax optimization problem which minimizes a group-structured optimal transport cost. Specifically, we focus on the applications of GSAT for group-sparse and rank-constrained perturbations modeled using group and nuclear norm penalties. In order to solve GSAT's non-smooth optimization problem in those cases, we propose a new minimax optimization algorithm called GDADMM by combining Gradient Descent Ascent (GDA) and Alternating Direction Method of Multipliers (ADMM). We present several applications of the GSAT framework to gain robustness against structured perturbations for image recognition and computational biology datasets." @default.
- W3175596306 created "2021-07-05" @default.
- W3175596306 creator A5005779176 @default.
- W3175596306 creator A5017160178 @default.
- W3175596306 creator A5024072566 @default.
- W3175596306 creator A5062650028 @default.
- W3175596306 date "2021-06-18" @default.
- W3175596306 modified "2023-09-27" @default.
- W3175596306 title "Group-Structured Adversarial Training." @default.
- W3175596306 cites W1945616565 @default.
- W3175596306 cites W2030611346 @default.
- W3175596306 cites W2049446938 @default.
- W3175596306 cites W2058401000 @default.
- W3175596306 cites W2062094438 @default.
- W3175596306 cites W2076626169 @default.
- W3175596306 cites W2095195675 @default.
- W3175596306 cites W2116200501 @default.
- W3175596306 cites W2118550318 @default.
- W3175596306 cites W2122825543 @default.
- W3175596306 cites W2135046866 @default.
- W3175596306 cites W2138019504 @default.
- W3175596306 cites W2163605009 @default.
- W3175596306 cites W2164278908 @default.
- W3175596306 cites W2176412452 @default.
- W3175596306 cites W2243397390 @default.
- W3175596306 cites W2253609413 @default.
- W3175596306 cites W2271840356 @default.
- W3175596306 cites W2296319761 @default.
- W3175596306 cites W2309685155 @default.
- W3175596306 cites W2329659234 @default.
- W3175596306 cites W2517229335 @default.
- W3175596306 cites W2543927648 @default.
- W3175596306 cites W2611328865 @default.
- W3175596306 cites W2618098489 @default.
- W3175596306 cites W2766972025 @default.
- W3175596306 cites W2786163515 @default.
- W3175596306 cites W2887906331 @default.
- W3175596306 cites W2891229686 @default.
- W3175596306 cites W2950687935 @default.
- W3175596306 cites W2963184668 @default.
- W3175596306 cites W2963274426 @default.
- W3175596306 cites W2963389226 @default.
- W3175596306 cites W2963450292 @default.
- W3175596306 cites W2963496101 @default.
- W3175596306 cites W2963539647 @default.
- W3175596306 cites W2963612069 @default.
- W3175596306 cites W2963744840 @default.
- W3175596306 cites W2963857521 @default.
- W3175596306 cites W2964110122 @default.
- W3175596306 cites W2964153729 @default.
- W3175596306 cites W2964253222 @default.
- W3175596306 cites W2970390122 @default.
- W3175596306 cites W2996252089 @default.
- W3175596306 cites W2996629283 @default.
- W3175596306 cites W2997502936 @default.
- W3175596306 cites W2998293245 @default.
- W3175596306 cites W3018690684 @default.
- W3175596306 cites W3029730229 @default.
- W3175596306 cites W3034862928 @default.
- W3175596306 cites W3037906261 @default.
- W3175596306 cites W3103340107 @default.
- W3175596306 cites W3104218734 @default.
- W3175596306 cites W3118608800 @default.
- W3175596306 cites W9657784 @default.
- W3175596306 hasPublicationYear "2021" @default.
- W3175596306 type Work @default.
- W3175596306 sameAs 3175596306 @default.
- W3175596306 citedByCount "0" @default.
- W3175596306 crossrefType "posted-content" @default.
- W3175596306 hasAuthorship W3175596306A5005779176 @default.
- W3175596306 hasAuthorship W3175596306A5017160178 @default.
- W3175596306 hasAuthorship W3175596306A5024072566 @default.
- W3175596306 hasAuthorship W3175596306A5062650028 @default.
- W3175596306 hasConcept C104317684 @default.
- W3175596306 hasConcept C11413529 @default.
- W3175596306 hasConcept C126255220 @default.
- W3175596306 hasConcept C137836250 @default.
- W3175596306 hasConcept C149728462 @default.
- W3175596306 hasConcept C153258448 @default.
- W3175596306 hasConcept C154945302 @default.
- W3175596306 hasConcept C185592680 @default.
- W3175596306 hasConcept C193254401 @default.
- W3175596306 hasConcept C33923547 @default.
- W3175596306 hasConcept C37736160 @default.
- W3175596306 hasConcept C41008148 @default.
- W3175596306 hasConcept C50644808 @default.
- W3175596306 hasConcept C55493867 @default.
- W3175596306 hasConcept C63479239 @default.
- W3175596306 hasConceptScore W3175596306C104317684 @default.
- W3175596306 hasConceptScore W3175596306C11413529 @default.
- W3175596306 hasConceptScore W3175596306C126255220 @default.
- W3175596306 hasConceptScore W3175596306C137836250 @default.
- W3175596306 hasConceptScore W3175596306C149728462 @default.
- W3175596306 hasConceptScore W3175596306C153258448 @default.
- W3175596306 hasConceptScore W3175596306C154945302 @default.
- W3175596306 hasConceptScore W3175596306C185592680 @default.
- W3175596306 hasConceptScore W3175596306C193254401 @default.
- W3175596306 hasConceptScore W3175596306C33923547 @default.
- W3175596306 hasConceptScore W3175596306C37736160 @default.
- W3175596306 hasConceptScore W3175596306C41008148 @default.