pcbi.1011792.s008.pdf (234.31 kB)
ED varies with model and training parameters.
journal contribution
posted on 2024-01-10, 18:35 authored by Eric Elmoznino, Michael F. BonnerTrends in ED across factors of variation in our models, such as layer depth and pre-training dataset.
(PDF)
History
Usage metrics
Categories
Keywords
representations onto lowpredicting cortical responsespopular view holdsnatural image representationslearning new categorieshuman fmri datalower dimensional geometrieshigher dimensional representationsvisual cortex benefitoptimal dnns compressdeep neural networksbetter generalization performancevisual cortexneural networksbetter performancedimensional subspacesbetter suitedbetter modelstraining domainsstrong trendmonkey electrophysiologylatent dimensionalitygeneralize beyondfindings suggestdnn modelscomputational modelsachieve invariance
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC