Matches in SemOpenAlex for { <https://semopenalex.org/work/W3166206819> ?p ?o ?g. }
Showing items 1 to 99 of
99
with 100 items per page.
- W3166206819 abstract "Deep neural networks are powerful machines for visual pattern recognition, but reasoning tasks that are easy for humans may still be difficult for neural models. Humans possess the ability to extrapolate reasoning strategies learned on simple problems to solve harder examples, often by for For example, a person who has learned to solve small mazes can easily extend the very same search techniques to solve much larger mazes by spending more time. In computers, this behavior is often achieved through the use of algorithms, which scale to arbitrarily hard problem instances at the cost of more computation. In contrast, the sequential computing budget of feed-forward neural networks is limited by their depth, and networks trained on simple problems have no way of extending their reasoning to accommodate harder problems. In this work, we show that recurrent networks trained to solve simple problems with few recurrent steps can indeed solve much more complex problems simply by performing additional recurrences during inference. We demonstrate this algorithmic behavior of recurrent networks on prefix sum computation, mazes, and chess. In all three domains, networks trained on simple problem instances are able to extend their reasoning abilities at test time simply by thinking for longer." @default.
- W3166206819 created "2021-06-22" @default.
- W3166206819 creator A5013018127 @default.
- W3166206819 creator A5014238711 @default.
- W3166206819 creator A5026712341 @default.
- W3166206819 creator A5037162168 @default.
- W3166206819 creator A5060687985 @default.
- W3166206819 creator A5066564672 @default.
- W3166206819 creator A5083662221 @default.
- W3166206819 date "2021-06-08" @default.
- W3166206819 modified "2023-09-23" @default.
- W3166206819 title "Can You Learn an Algorithm? Generalizing from Easy to Hard Problems with Recurrent Networks" @default.
- W3166206819 cites W1546771929 @default.
- W3166206819 cites W1597864774 @default.
- W3166206819 cites W1934184906 @default.
- W3166206819 cites W2096707172 @default.
- W3166206819 cites W2173051530 @default.
- W3166206819 cites W2194775991 @default.
- W3166206819 cites W2258731934 @default.
- W3166206819 cites W2293709543 @default.
- W3166206819 cites W2325237720 @default.
- W3166206819 cites W2331143823 @default.
- W3166206819 cites W2337199865 @default.
- W3166206819 cites W2772709170 @default.
- W3166206819 cites W2786776430 @default.
- W3166206819 cites W2788906943 @default.
- W3166206819 cites W2949512190 @default.
- W3166206819 cites W2950527759 @default.
- W3166206819 cites W2951244744 @default.
- W3166206819 cites W2953494151 @default.
- W3166206819 cites W2963016848 @default.
- W3166206819 cites W2970900903 @default.
- W3166206819 cites W2996428491 @default.
- W3166206819 cites W3034846572 @default.
- W3166206819 cites W3087471160 @default.
- W3166206819 cites W3099168273 @default.
- W3166206819 cites W3134144764 @default.
- W3166206819 cites W3194752246 @default.
- W3166206819 hasPublicationYear "2021" @default.
- W3166206819 type Work @default.
- W3166206819 sameAs 3166206819 @default.
- W3166206819 citedByCount "0" @default.
- W3166206819 crossrefType "posted-content" @default.
- W3166206819 hasAuthorship W3166206819A5013018127 @default.
- W3166206819 hasAuthorship W3166206819A5014238711 @default.
- W3166206819 hasAuthorship W3166206819A5026712341 @default.
- W3166206819 hasAuthorship W3166206819A5037162168 @default.
- W3166206819 hasAuthorship W3166206819A5060687985 @default.
- W3166206819 hasAuthorship W3166206819A5066564672 @default.
- W3166206819 hasAuthorship W3166206819A5083662221 @default.
- W3166206819 hasConcept C111472728 @default.
- W3166206819 hasConcept C11413529 @default.
- W3166206819 hasConcept C119857082 @default.
- W3166206819 hasConcept C138885662 @default.
- W3166206819 hasConcept C147168706 @default.
- W3166206819 hasConcept C154945302 @default.
- W3166206819 hasConcept C2776214188 @default.
- W3166206819 hasConcept C2780586882 @default.
- W3166206819 hasConcept C41008148 @default.
- W3166206819 hasConcept C45374587 @default.
- W3166206819 hasConcept C50644808 @default.
- W3166206819 hasConceptScore W3166206819C111472728 @default.
- W3166206819 hasConceptScore W3166206819C11413529 @default.
- W3166206819 hasConceptScore W3166206819C119857082 @default.
- W3166206819 hasConceptScore W3166206819C138885662 @default.
- W3166206819 hasConceptScore W3166206819C147168706 @default.
- W3166206819 hasConceptScore W3166206819C154945302 @default.
- W3166206819 hasConceptScore W3166206819C2776214188 @default.
- W3166206819 hasConceptScore W3166206819C2780586882 @default.
- W3166206819 hasConceptScore W3166206819C41008148 @default.
- W3166206819 hasConceptScore W3166206819C45374587 @default.
- W3166206819 hasConceptScore W3166206819C50644808 @default.
- W3166206819 hasLocation W31662068191 @default.
- W3166206819 hasOpenAccess W3166206819 @default.
- W3166206819 hasPrimaryLocation W31662068191 @default.
- W3166206819 hasRelatedWork W1581508004 @default.
- W3166206819 hasRelatedWork W1875573847 @default.
- W3166206819 hasRelatedWork W2189255447 @default.
- W3166206819 hasRelatedWork W2476674146 @default.
- W3166206819 hasRelatedWork W2521274174 @default.
- W3166206819 hasRelatedWork W2563734883 @default.
- W3166206819 hasRelatedWork W2613603362 @default.
- W3166206819 hasRelatedWork W2772251499 @default.
- W3166206819 hasRelatedWork W2805983332 @default.
- W3166206819 hasRelatedWork W2895551046 @default.
- W3166206819 hasRelatedWork W2963929497 @default.
- W3166206819 hasRelatedWork W3008124487 @default.
- W3166206819 hasRelatedWork W3025133903 @default.
- W3166206819 hasRelatedWork W3054093752 @default.
- W3166206819 hasRelatedWork W3100283536 @default.
- W3166206819 hasRelatedWork W3118743894 @default.
- W3166206819 hasRelatedWork W3129990728 @default.
- W3166206819 hasRelatedWork W3160733000 @default.
- W3166206819 hasRelatedWork W3171374279 @default.
- W3166206819 hasRelatedWork W3213671677 @default.
- W3166206819 isParatext "false" @default.
- W3166206819 isRetracted "false" @default.
- W3166206819 magId "3166206819" @default.
- W3166206819 workType "article" @default.