Matches in SemOpenAlex for { <https://semopenalex.org/work/W4225526255> ?p ?o ?g. }
Showing items 1 to 77 of
77
with 100 items per page.
- W4225526255 abstract "It has been reported that deep learning models are extremely vulnerable to small but intentionally chosen perturbations of its input. In particular, a deep network, despite its near-optimal accuracy on the clean images, often mis-classifies an image with a worst-case but humanly imperceptible perturbation (so-called adversarial examples). To tackle this problem, a great amount of research has been done to study the training procedure of a network to improve its robustness. However, most of the research so far has focused on the case of supervised learning. With the increasing popularity of self-supervised learning methods, it is also important to study and improve the robustness of their resulting representation on the downstream tasks. In this paper, we study the problem of robust representation learning with unlabeled data in a task-agnostic manner. Specifically, we first derive an upper bound on the adversarial loss of a prediction model (which is based on the learned representation) on any downstream task, using its loss on the clean data and a robustness regularizer. Moreover, the regularizer is task-independent, thus we propose to minimize it directly during the representation learning phase to make the downstream prediction model more robust. Extensive experiments show that our method achieves preferable adversarial performance compared to relevant baselines." @default.
- W4225526255 created "2022-05-05" @default.
- W4225526255 creator A5012529382 @default.
- W4225526255 creator A5067605731 @default.
- W4225526255 creator A5087829140 @default.
- W4225526255 date "2022-03-14" @default.
- W4225526255 modified "2023-10-16" @default.
- W4225526255 title "Task-Agnostic Robust Representation Learning" @default.
- W4225526255 doi "https://doi.org/10.48550/arxiv.2203.07596" @default.
- W4225526255 hasPublicationYear "2022" @default.
- W4225526255 type Work @default.
- W4225526255 citedByCount "0" @default.
- W4225526255 crossrefType "posted-content" @default.
- W4225526255 hasAuthorship W4225526255A5012529382 @default.
- W4225526255 hasAuthorship W4225526255A5067605731 @default.
- W4225526255 hasAuthorship W4225526255A5087829140 @default.
- W4225526255 hasBestOaLocation W42255262551 @default.
- W4225526255 hasConcept C104317684 @default.
- W4225526255 hasConcept C108583219 @default.
- W4225526255 hasConcept C119857082 @default.
- W4225526255 hasConcept C154945302 @default.
- W4225526255 hasConcept C15744967 @default.
- W4225526255 hasConcept C162324750 @default.
- W4225526255 hasConcept C17744445 @default.
- W4225526255 hasConcept C185592680 @default.
- W4225526255 hasConcept C187736073 @default.
- W4225526255 hasConcept C199539241 @default.
- W4225526255 hasConcept C2776145971 @default.
- W4225526255 hasConcept C2776359362 @default.
- W4225526255 hasConcept C2780451532 @default.
- W4225526255 hasConcept C2780586970 @default.
- W4225526255 hasConcept C37736160 @default.
- W4225526255 hasConcept C41008148 @default.
- W4225526255 hasConcept C51632099 @default.
- W4225526255 hasConcept C55493867 @default.
- W4225526255 hasConcept C59404180 @default.
- W4225526255 hasConcept C63479239 @default.
- W4225526255 hasConcept C77805123 @default.
- W4225526255 hasConcept C94625758 @default.
- W4225526255 hasConceptScore W4225526255C104317684 @default.
- W4225526255 hasConceptScore W4225526255C108583219 @default.
- W4225526255 hasConceptScore W4225526255C119857082 @default.
- W4225526255 hasConceptScore W4225526255C154945302 @default.
- W4225526255 hasConceptScore W4225526255C15744967 @default.
- W4225526255 hasConceptScore W4225526255C162324750 @default.
- W4225526255 hasConceptScore W4225526255C17744445 @default.
- W4225526255 hasConceptScore W4225526255C185592680 @default.
- W4225526255 hasConceptScore W4225526255C187736073 @default.
- W4225526255 hasConceptScore W4225526255C199539241 @default.
- W4225526255 hasConceptScore W4225526255C2776145971 @default.
- W4225526255 hasConceptScore W4225526255C2776359362 @default.
- W4225526255 hasConceptScore W4225526255C2780451532 @default.
- W4225526255 hasConceptScore W4225526255C2780586970 @default.
- W4225526255 hasConceptScore W4225526255C37736160 @default.
- W4225526255 hasConceptScore W4225526255C41008148 @default.
- W4225526255 hasConceptScore W4225526255C51632099 @default.
- W4225526255 hasConceptScore W4225526255C55493867 @default.
- W4225526255 hasConceptScore W4225526255C59404180 @default.
- W4225526255 hasConceptScore W4225526255C63479239 @default.
- W4225526255 hasConceptScore W4225526255C77805123 @default.
- W4225526255 hasConceptScore W4225526255C94625758 @default.
- W4225526255 hasLocation W42255262551 @default.
- W4225526255 hasOpenAccess W4225526255 @default.
- W4225526255 hasPrimaryLocation W42255262551 @default.
- W4225526255 hasRelatedWork W2610321374 @default.
- W4225526255 hasRelatedWork W2952919291 @default.
- W4225526255 hasRelatedWork W2966207284 @default.
- W4225526255 hasRelatedWork W3174891612 @default.
- W4225526255 hasRelatedWork W3193857078 @default.
- W4225526255 hasRelatedWork W4225526255 @default.
- W4225526255 hasRelatedWork W4226213069 @default.
- W4225526255 hasRelatedWork W4287116584 @default.
- W4225526255 hasRelatedWork W4287777115 @default.
- W4225526255 hasRelatedWork W4286890323 @default.
- W4225526255 isParatext "false" @default.
- W4225526255 isRetracted "false" @default.
- W4225526255 workType "article" @default.