Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287064260> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4287064260 abstract "Deep neural networks set the state-of-the-art across many tasks in computer vision, but their generalization ability to image distortions is surprisingly fragile. In contrast, the mammalian visual system is robust to a wide range of perturbations. Recent work suggests that this generalization ability can be explained by useful inductive biases encoded in the representations of visual stimuli throughout the visual cortex. Here, we successfully leveraged these inductive biases with a multi-task learning approach: we jointly trained a deep network to perform image classification and to predict neural activity in macaque primary visual cortex (V1). We measured the out-of-distribution generalization abilities of our network by testing its robustness to image distortions. We found that co-training on monkey V1 data leads to increased robustness despite the absence of those distortions during training. Additionally, we showed that our network's robustness is very close to that of an Oracle network where parts of the architecture are directly trained on noisy images. Our results also demonstrated that the network's representations become more brain-like as their robustness improves. Using a novel constrained reconstruction analysis, we investigated what makes our brain-regularized network more robust. We found that our co-trained network is more sensitive to content than noise when compared to a Baseline network that we trained for image classification alone. Using DeepGaze-predicted saliency maps for ImageNet images, we found that our monkey co-trained network tends to be more sensitive to salient regions in a scene, reminiscent of existing theories on the role of V1 in the detection of object borders and bottom-up saliency. Overall, our work expands the promising research avenue of transferring inductive biases from the brain, and provides a novel analysis of the effects of our transfer." @default.
- W4287064260 created "2022-07-25" @default.
- W4287064260 creator A5019844889 @default.
- W4287064260 creator A5021706641 @default.
- W4287064260 creator A5024493026 @default.
- W4287064260 creator A5036361185 @default.
- W4287064260 creator A5054637446 @default.
- W4287064260 creator A5057996614 @default.
- W4287064260 creator A5077061252 @default.
- W4287064260 creator A5079206365 @default.
- W4287064260 date "2021-07-29" @default.
- W4287064260 modified "2023-09-25" @default.
- W4287064260 title "Towards robust vision by multi-task learning on monkey visual cortex" @default.
- W4287064260 doi "https://doi.org/10.48550/arxiv.2107.14344" @default.
- W4287064260 hasPublicationYear "2021" @default.
- W4287064260 type Work @default.
- W4287064260 citedByCount "0" @default.
- W4287064260 crossrefType "posted-content" @default.
- W4287064260 hasAuthorship W4287064260A5019844889 @default.
- W4287064260 hasAuthorship W4287064260A5021706641 @default.
- W4287064260 hasAuthorship W4287064260A5024493026 @default.
- W4287064260 hasAuthorship W4287064260A5036361185 @default.
- W4287064260 hasAuthorship W4287064260A5054637446 @default.
- W4287064260 hasAuthorship W4287064260A5057996614 @default.
- W4287064260 hasAuthorship W4287064260A5077061252 @default.
- W4287064260 hasAuthorship W4287064260A5079206365 @default.
- W4287064260 hasBestOaLocation W42870642601 @default.
- W4287064260 hasConcept C104317684 @default.
- W4287064260 hasConcept C119857082 @default.
- W4287064260 hasConcept C134306372 @default.
- W4287064260 hasConcept C153180895 @default.
- W4287064260 hasConcept C154945302 @default.
- W4287064260 hasConcept C15744967 @default.
- W4287064260 hasConcept C169760540 @default.
- W4287064260 hasConcept C177148314 @default.
- W4287064260 hasConcept C2779345533 @default.
- W4287064260 hasConcept C31972630 @default.
- W4287064260 hasConcept C33923547 @default.
- W4287064260 hasConcept C41008148 @default.
- W4287064260 hasConcept C50644808 @default.
- W4287064260 hasConcept C55493867 @default.
- W4287064260 hasConcept C63479239 @default.
- W4287064260 hasConcept C86803240 @default.
- W4287064260 hasConceptScore W4287064260C104317684 @default.
- W4287064260 hasConceptScore W4287064260C119857082 @default.
- W4287064260 hasConceptScore W4287064260C134306372 @default.
- W4287064260 hasConceptScore W4287064260C153180895 @default.
- W4287064260 hasConceptScore W4287064260C154945302 @default.
- W4287064260 hasConceptScore W4287064260C15744967 @default.
- W4287064260 hasConceptScore W4287064260C169760540 @default.
- W4287064260 hasConceptScore W4287064260C177148314 @default.
- W4287064260 hasConceptScore W4287064260C2779345533 @default.
- W4287064260 hasConceptScore W4287064260C31972630 @default.
- W4287064260 hasConceptScore W4287064260C33923547 @default.
- W4287064260 hasConceptScore W4287064260C41008148 @default.
- W4287064260 hasConceptScore W4287064260C50644808 @default.
- W4287064260 hasConceptScore W4287064260C55493867 @default.
- W4287064260 hasConceptScore W4287064260C63479239 @default.
- W4287064260 hasConceptScore W4287064260C86803240 @default.
- W4287064260 hasLocation W42870642601 @default.
- W4287064260 hasOpenAccess W4287064260 @default.
- W4287064260 hasPrimaryLocation W42870642601 @default.
- W4287064260 hasRelatedWork W1999222583 @default.
- W4287064260 hasRelatedWork W2035976912 @default.
- W4287064260 hasRelatedWork W2042311553 @default.
- W4287064260 hasRelatedWork W2109974539 @default.
- W4287064260 hasRelatedWork W2125927971 @default.
- W4287064260 hasRelatedWork W2146875012 @default.
- W4287064260 hasRelatedWork W2541791370 @default.
- W4287064260 hasRelatedWork W2738084969 @default.
- W4287064260 hasRelatedWork W2954664659 @default.
- W4287064260 hasRelatedWork W2980357211 @default.
- W4287064260 isParatext "false" @default.
- W4287064260 isRetracted "false" @default.
- W4287064260 workType "article" @default.