Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386857325> ?p ?o ?g. }
Showing items 1 to 74 of
74
with 100 items per page.
- W4386857325 endingPage "1869" @default.
- W4386857325 startingPage "1850" @default.
- W4386857325 abstract "Recurrent neural networks (RNNs) are often used to model circuits in the brain and can solve a variety of difficult computational problems requiring memory, error correction, or selection (Hopfield, 1982; Maass et al., 2002; Maass, 2011). However, fully connected RNNs contrast structurally with their biological counterparts, which are extremely sparse (about 0.1%). Motivated by the neocortex, where neural connectivity is constrained by physical distance along cortical sheets and other synaptic wiring costs, we introduce locality masked RNNs (LM-RNNs) that use task-agnostic predetermined graphs with sparsity as low as 4%. We study LM-RNNs in a multitask learning setting relevant to cognitive systems neuroscience with a commonly used set of tasks, 20-Cog-tasks (Yang et al., 2019). We show through reductio ad absurdum that 20-Cog-tasks can be solved by a small pool of separated autapses that we can mechanistically analyze and understand. Thus, these tasks fall short of the goal of inducing complex recurrent dynamics and modular structure in RNNs. We next contribute a new cognitive multitask battery, Mod-Cog, consisting of up to 132 tasks that expands by about seven-fold the number of tasks and task complexity of 20-Cog-tasks. Importantly, while autapses can solve the simple 20-Cog-tasks, the expanded task set requires richer neural architectures and continuous attractor dynamics. On these tasks, we show that LM-RNNs with an optimal sparsity result in faster training and better data efficiency than fully connected networks." @default.
- W4386857325 created "2023-09-20" @default.
- W4386857325 creator A5025224995 @default.
- W4386857325 creator A5057755749 @default.
- W4386857325 creator A5068834876 @default.
- W4386857325 creator A5083541135 @default.
- W4386857325 date "2023-10-10" @default.
- W4386857325 modified "2023-10-18" @default.
- W4386857325 title "Winning the Lottery With Neural Connectivity Constraints: Faster Learning Across Cognitive Tasks With Spatially Constrained Sparse RNNs" @default.
- W4386857325 cites W1993665893 @default.
- W4386857325 cites W2103179919 @default.
- W4386857325 cites W2112090702 @default.
- W4386857325 cites W2128084896 @default.
- W4386857325 cites W2167845354 @default.
- W4386857325 cites W2171865010 @default.
- W4386857325 cites W2337788193 @default.
- W4386857325 cites W2371946849 @default.
- W4386857325 cites W2588048611 @default.
- W4386857325 cites W2908124316 @default.
- W4386857325 cites W3018844151 @default.
- W4386857325 cites W3184481189 @default.
- W4386857325 cites W3198296750 @default.
- W4386857325 cites W4210958102 @default.
- W4386857325 cites W4280533591 @default.
- W4386857325 cites W4308294084 @default.
- W4386857325 doi "https://doi.org/10.1162/neco_a_01613" @default.
- W4386857325 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37725708" @default.
- W4386857325 hasPublicationYear "2023" @default.
- W4386857325 type Work @default.
- W4386857325 citedByCount "0" @default.
- W4386857325 crossrefType "journal-article" @default.
- W4386857325 hasAuthorship W4386857325A5025224995 @default.
- W4386857325 hasAuthorship W4386857325A5057755749 @default.
- W4386857325 hasAuthorship W4386857325A5068834876 @default.
- W4386857325 hasAuthorship W4386857325A5083541135 @default.
- W4386857325 hasConcept C119857082 @default.
- W4386857325 hasConcept C147168706 @default.
- W4386857325 hasConcept C154945302 @default.
- W4386857325 hasConcept C162324750 @default.
- W4386857325 hasConcept C187736073 @default.
- W4386857325 hasConcept C190502265 @default.
- W4386857325 hasConcept C2780451532 @default.
- W4386857325 hasConcept C41008148 @default.
- W4386857325 hasConcept C50644808 @default.
- W4386857325 hasConceptScore W4386857325C119857082 @default.
- W4386857325 hasConceptScore W4386857325C147168706 @default.
- W4386857325 hasConceptScore W4386857325C154945302 @default.
- W4386857325 hasConceptScore W4386857325C162324750 @default.
- W4386857325 hasConceptScore W4386857325C187736073 @default.
- W4386857325 hasConceptScore W4386857325C190502265 @default.
- W4386857325 hasConceptScore W4386857325C2780451532 @default.
- W4386857325 hasConceptScore W4386857325C41008148 @default.
- W4386857325 hasConceptScore W4386857325C50644808 @default.
- W4386857325 hasIssue "11" @default.
- W4386857325 hasLocation W43868573251 @default.
- W4386857325 hasLocation W43868573252 @default.
- W4386857325 hasOpenAccess W4386857325 @default.
- W4386857325 hasPrimaryLocation W43868573251 @default.
- W4386857325 hasRelatedWork W2590796488 @default.
- W4386857325 hasRelatedWork W2734358244 @default.
- W4386857325 hasRelatedWork W2750384547 @default.
- W4386857325 hasRelatedWork W2809732489 @default.
- W4386857325 hasRelatedWork W2886711096 @default.
- W4386857325 hasRelatedWork W2950475743 @default.
- W4386857325 hasRelatedWork W3046591097 @default.
- W4386857325 hasRelatedWork W3088091256 @default.
- W4386857325 hasRelatedWork W4380078352 @default.
- W4386857325 hasRelatedWork W4386603768 @default.
- W4386857325 hasVolume "35" @default.
- W4386857325 isParatext "false" @default.
- W4386857325 isRetracted "false" @default.
- W4386857325 workType "article" @default.