FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2996170846> ?p ?o ?g. }

• W2996170846 abstract "Recurrent neural networks (RNNs) provide a powerful tool for online prediction in online partially observable problems. However, there are two primary issues one must overcome when training an RNN: the sensitivity of the learning algorithm’s performance to truncation length and and long training times. There are variety of strategies to improve training in RNNs, particularly with Backprop Through Time (BPTT) and by Real-Time Recurrent Learning. These strategies, however, are typically computationally expensive and focus computation on computing gradients back in time. In this work, we reformulate the RNN training objective to explicitly learn state vectors; this breaks the dependence across time and so avoids the need to estimate gradients far back in time. We show that for a fixed buffer of data, our algorithm—called Fixed Point Propagation (FPP)—is sound: it converges to a stationary point of the new objective. We investigate the empirical performance of our online FPP algorithm, particularly in terms of computation compared to truncated BPTT with varying truncation levels." @default.
• W2996170846 created "2019-12-26" @default.
• W2996170846 creator A5018071900 @default.
• W2996170846 creator A5031606152 @default.
• W2996170846 creator A5050561161 @default.
• W2996170846 creator A5071842426 @default.
• W2996170846 creator A5081160595 @default.
• W2996170846 date "2020-04-30" @default.
• W2996170846 modified "2023-09-23" @default.
• W2996170846 title "Training Recurrent Neural Networks Online by Learning Explicit State Variables" @default.
• W2996170846 cites W1570106308 @default.
• W2996170846 cites W1669104078 @default.
• W2996170846 cites W1815076433 @default.
• W2996170846 cites W1881179843 @default.
• W2996170846 cites W2007431958 @default.
• W2996170846 cites W2016589492 @default.
• W2996170846 cites W2026641456 @default.
• W2996170846 cites W2029463628 @default.
• W2996170846 cites W2057653135 @default.
• W2996170846 cites W2064675550 @default.
• W2996170846 cites W2110485445 @default.
• W2996170846 cites W2128084896 @default.
• W2996170846 cites W2143612262 @default.
• W2996170846 cites W2150355110 @default.
• W2996170846 cites W2154890045 @default.
• W2996170846 cites W2159243025 @default.
• W2996170846 cites W2160815625 @default.
• W2996170846 cites W2327501763 @default.
• W2996170846 cites W2346438296 @default.
• W2996170846 cites W2510842514 @default.
• W2996170846 cites W2547418827 @default.
• W2996170846 cites W2584032004 @default.
• W2996170846 cites W2749590927 @default.
• W2996170846 cites W2899217980 @default.
• W2996170846 cites W2908064123 @default.
• W2996170846 cites W2910812475 @default.
• W2996170846 cites W2952469094 @default.
• W2996170846 cites W2963042606 @default.
• W2996170846 cites W2963055445 @default.
• W2996170846 cites W2963140169 @default.
• W2996170846 cites W2963211739 @default.
• W2996170846 cites W2963248893 @default.
• W2996170846 cites W2964243329 @default.
• W2996170846 cites W2964330624 @default.
• W2996170846 hasPublicationYear "2020" @default.
• W2996170846 type Work @default.
• W2996170846 sameAs 2996170846 @default.
• W2996170846 citedByCount "0" @default.
• W2996170846 crossrefType "proceedings-article" @default.
• W2996170846 hasAuthorship W2996170846A5018071900 @default.
• W2996170846 hasAuthorship W2996170846A5031606152 @default.
• W2996170846 hasAuthorship W2996170846A5050561161 @default.
• W2996170846 hasAuthorship W2996170846A5071842426 @default.
• W2996170846 hasAuthorship W2996170846A5081160595 @default.
• W2996170846 hasConcept C106195933 @default.
• W2996170846 hasConcept C11413529 @default.
• W2996170846 hasConcept C119857082 @default.
• W2996170846 hasConcept C120665830 @default.
• W2996170846 hasConcept C121332964 @default.
• W2996170846 hasConcept C126255220 @default.
• W2996170846 hasConcept C136197465 @default.
• W2996170846 hasConcept C136764020 @default.
• W2996170846 hasConcept C147168706 @default.
• W2996170846 hasConcept C154945302 @default.
• W2996170846 hasConcept C192209626 @default.
• W2996170846 hasConcept C2986087404 @default.
• W2996170846 hasConcept C33923547 @default.
• W2996170846 hasConcept C41008148 @default.
• W2996170846 hasConcept C45374587 @default.
• W2996170846 hasConcept C50644808 @default.
• W2996170846 hasConceptScore W2996170846C106195933 @default.
• W2996170846 hasConceptScore W2996170846C11413529 @default.
• W2996170846 hasConceptScore W2996170846C119857082 @default.
• W2996170846 hasConceptScore W2996170846C120665830 @default.
• W2996170846 hasConceptScore W2996170846C121332964 @default.
• W2996170846 hasConceptScore W2996170846C126255220 @default.
• W2996170846 hasConceptScore W2996170846C136197465 @default.
• W2996170846 hasConceptScore W2996170846C136764020 @default.
• W2996170846 hasConceptScore W2996170846C147168706 @default.
• W2996170846 hasConceptScore W2996170846C154945302 @default.
• W2996170846 hasConceptScore W2996170846C192209626 @default.
• W2996170846 hasConceptScore W2996170846C2986087404 @default.
• W2996170846 hasConceptScore W2996170846C33923547 @default.
• W2996170846 hasConceptScore W2996170846C41008148 @default.
• W2996170846 hasConceptScore W2996170846C45374587 @default.
• W2996170846 hasConceptScore W2996170846C50644808 @default.
• W2996170846 hasLocation W29961708461 @default.
• W2996170846 hasOpenAccess W2996170846 @default.
• W2996170846 hasPrimaryLocation W29961708461 @default.
• W2996170846 hasRelatedWork W136010251 @default.
• W2996170846 hasRelatedWork W1523116777 @default.
• W2996170846 hasRelatedWork W1594702668 @default.
• W2996170846 hasRelatedWork W2029897006 @default.
• W2996170846 hasRelatedWork W2116322539 @default.
• W2996170846 hasRelatedWork W2128712546 @default.
• W2996170846 hasRelatedWork W2252143850 @default.
• W2996170846 hasRelatedWork W2408741041 @default.
• W2996170846 hasRelatedWork W2737836580 @default.
• W2996170846 hasRelatedWork W2804407542 @default.
• W2996170846 hasRelatedWork W2863476589 @default.

• next