FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3146366147> ?p ?o ?g. }

• W3146366147 abstract "Abstract Learning from a limited number of experiences requires suitable inductive biases. To identify how inductive biases are implemented in and shaped by neural codes, we analyze sample-efficient learning of arbitrary stimulus-response maps from arbitrary neural codes with biologically-plausible readouts. We develop an analytical theory that predicts the generalization error of the readout as a function of the number of observed examples. Our theory illustrates in a mathematically precise way how the structure of population codes shapes inductive bias, and how a match between the code and the task is crucial for sample-efficient learning. We observe that many different codes can support the same inductive bias. By analyzing recordings from the mouse primary visual cortex, we demonstrate that biological codes have lower total activity than other codes with identical bias. Using these mouse primary visual cortex responses, we demonstrate the existence of an efficiency bias towards low frequency orientation discrimination tasks for grating stimuli and low spatial frequency reconstruction tasks for natural images. We reproduce the discrimination bias in a simple model of primary visual cortex, and further show how invariances in the code to certain stimulus variations alter learning performance. We extend our methods to time-dependent neural codes and predict the sample efficiency of readouts from recurrent networks. Finally, we discuss implications of our theory in the context of recent developments in neuroscience and artificial intelligence. Overall, our study provides a concrete method for elucidating inductive biases of the brain and promotes sample-efficient learning as a general normative coding principle." @default.
• W3146366147 created "2021-04-13" @default.
• W3146366147 creator A5023195984 @default.
• W3146366147 creator A5039282308 @default.
• W3146366147 date "2021-03-31" @default.
• W3146366147 modified "2023-09-27" @default.
• W3146366147 title "Population codes enable learning from few examples by shaping inductive bias" @default.
• W3146366147 cites W1535810436 @default.
• W3146366147 cites W1540155273 @default.
• W3146366147 cites W1650706190 @default.
• W3146366147 cites W1868105705 @default.
• W3146366147 cites W1961056875 @default.
• W3146366147 cites W1963769676 @default.
• W3146366147 cites W1966537851 @default.
• W3146366147 cites W1968942430 @default.
• W3146366147 cites W1969469858 @default.
• W3146366147 cites W1981025032 @default.
• W3146366147 cites W1986098213 @default.
• W3146366147 cites W1998050452 @default.
• W3146366147 cites W1998462024 @default.
• W3146366147 cites W2009284521 @default.
• W3146366147 cites W2014158063 @default.
• W3146366147 cites W2017539895 @default.
• W3146366147 cites W2024969564 @default.
• W3146366147 cites W2026928404 @default.
• W3146366147 cites W2027802883 @default.
• W3146366147 cites W2029137641 @default.
• W3146366147 cites W2033705462 @default.
• W3146366147 cites W2036899314 @default.
• W3146366147 cites W2037843057 @default.
• W3146366147 cites W2042925217 @default.
• W3146366147 cites W2043850729 @default.
• W3146366147 cites W2049977058 @default.
• W3146366147 cites W2051526952 @default.
• W3146366147 cites W2057736028 @default.
• W3146366147 cites W2063632045 @default.
• W3146366147 cites W2074423218 @default.
• W3146366147 cites W2075067380 @default.
• W3146366147 cites W2077381770 @default.
• W3146366147 cites W2079960018 @default.
• W3146366147 cites W2086841290 @default.
• W3146366147 cites W2095087224 @default.
• W3146366147 cites W2100483895 @default.
• W3146366147 cites W2103314569 @default.
• W3146366147 cites W2105430025 @default.
• W3146366147 cites W2105464873 @default.
• W3146366147 cites W2108598243 @default.
• W3146366147 cites W2108992228 @default.
• W3146366147 cites W2111306669 @default.
• W3146366147 cites W2119885245 @default.
• W3146366147 cites W2122741244 @default.
• W3146366147 cites W2123713131 @default.
• W3146366147 cites W2140552585 @default.
• W3146366147 cites W2141927780 @default.
• W3146366147 cites W2151035455 @default.
• W3146366147 cites W2160996222 @default.
• W3146366147 cites W2164365056 @default.
• W3146366147 cites W2166149038 @default.
• W3146366147 cites W2167965189 @default.
• W3146366147 cites W2170697082 @default.
• W3146366147 cites W2171281863 @default.
• W3146366147 cites W2588048611 @default.
• W3146366147 cites W2592738714 @default.
• W3146366147 cites W2621961142 @default.
• W3146366147 cites W2735683573 @default.
• W3146366147 cites W2883327607 @default.
• W3146366147 cites W2885162916 @default.
• W3146366147 cites W2896846197 @default.
• W3146366147 cites W2949895299 @default.
• W3146366147 cites W2963305465 @default.
• W3146366147 cites W2963518130 @default.
• W3146366147 cites W2975512858 @default.
• W3146366147 cites W3004619146 @default.
• W3146366147 cites W3008714724 @default.
• W3146366147 cites W3016026822 @default.
• W3146366147 cites W3018252856 @default.
• W3146366147 cites W3162003518 @default.
• W3146366147 cites W3201017316 @default.
• W3146366147 cites W4211049957 @default.
• W3146366147 cites W4211189800 @default.
• W3146366147 cites W4214871172 @default.
• W3146366147 cites W4230449744 @default.
• W3146366147 cites W4240046814 @default.
• W3146366147 doi "https://doi.org/10.1101/2021.03.30.437743" @default.
• W3146366147 hasPublicationYear "2021" @default.
• W3146366147 type Work @default.
• W3146366147 sameAs 3146366147 @default.
• W3146366147 citedByCount "5" @default.
• W3146366147 countsByYear W31463661472021 @default.
• W3146366147 countsByYear W31463661472022 @default.
• W3146366147 crossrefType "posted-content" @default.
• W3146366147 hasAuthorship W3146366147A5023195984 @default.
• W3146366147 hasAuthorship W3146366147A5039282308 @default.
• W3146366147 hasBestOaLocation W31463661471 @default.
• W3146366147 hasConcept C105795698 @default.
• W3146366147 hasConcept C119857082 @default.
• W3146366147 hasConcept C144024400 @default.
• W3146366147 hasConcept C149923435 @default.
• W3146366147 hasConcept C153180895 @default.
• W3146366147 hasConcept C154945302 @default.

• next