Matches in SemOpenAlex for { <https://semopenalex.org/work/W3210368837> ?p ?o ?g. }
Showing items 1 to 81 of
81
with 100 items per page.
- W3210368837 endingPage "153" @default.
- W3210368837 startingPage "130" @default.
- W3210368837 abstract "Recurrent neural networks (RNNs) are a widely used deep architecture for sequence modeling, generation, and prediction. Despite success in applications such as machine translation and voice recognition, these stateful models have several critical shortcomings. Specifically, RNNs generalize poorly over very long sequences, which limits their applicability to many important temporal processing and time series forecasting problems. For example, RNNs struggle in recognizing complex context free languages (CFLs), never reaching 100% accuracy on training. One way to address these shortcomings is to couple an RNN with an external, differentiable memory structure, such as a stack. However, differentiable memories in prior work have neither been extensively studied on CFLs nor tested on sequences longer than those seen in training. The few efforts that have studied them have shown that continuous differentiable memory structures yield poor generalization for complex CFLs, making the RNN less interpretable. In this paper, we improve the memory-augmented RNN with important architectural and state updating mechanisms that ensure that the model learns to properly balance the use of its latent states with external memory. Our improved RNN models exhibit better generalization performance and are able to classify long strings generated by complex hierarchical context free grammars (CFGs). We evaluate our models on CGGs, including the Dyck languages, as well as on the Penn Treebank language modelling task, and achieve stable, robust performance across these benchmarks. Furthermore, we show that only our memory-augmented networks are capable of retaining memory for a longer duration up to strings of length 160." @default.
- W3210368837 created "2021-11-08" @default.
- W3210368837 creator A5001294898 @default.
- W3210368837 creator A5005431144 @default.
- W3210368837 creator A5068199296 @default.
- W3210368837 creator A5084332360 @default.
- W3210368837 date "2020-01-01" @default.
- W3210368837 modified "2023-10-14" @default.
- W3210368837 title "Recognizing Long Grammatical Sequences using Recurrent Networks Augmented with an External Differentiable Stack" @default.
- W3210368837 hasPublicationYear "2020" @default.
- W3210368837 type Work @default.
- W3210368837 sameAs 3210368837 @default.
- W3210368837 citedByCount "2" @default.
- W3210368837 countsByYear W32103688372020 @default.
- W3210368837 countsByYear W32103688372021 @default.
- W3210368837 crossrefType "journal-article" @default.
- W3210368837 hasAuthorship W3210368837A5001294898 @default.
- W3210368837 hasAuthorship W3210368837A5005431144 @default.
- W3210368837 hasAuthorship W3210368837A5068199296 @default.
- W3210368837 hasAuthorship W3210368837A5084332360 @default.
- W3210368837 hasConcept C119857082 @default.
- W3210368837 hasConcept C134306372 @default.
- W3210368837 hasConcept C137293760 @default.
- W3210368837 hasConcept C147168706 @default.
- W3210368837 hasConcept C151730666 @default.
- W3210368837 hasConcept C154945302 @default.
- W3210368837 hasConcept C177148314 @default.
- W3210368837 hasConcept C186644900 @default.
- W3210368837 hasConcept C202615002 @default.
- W3210368837 hasConcept C204321447 @default.
- W3210368837 hasConcept C206134035 @default.
- W3210368837 hasConcept C2779343474 @default.
- W3210368837 hasConcept C33923547 @default.
- W3210368837 hasConcept C41008148 @default.
- W3210368837 hasConcept C50644808 @default.
- W3210368837 hasConcept C86803240 @default.
- W3210368837 hasConceptScore W3210368837C119857082 @default.
- W3210368837 hasConceptScore W3210368837C134306372 @default.
- W3210368837 hasConceptScore W3210368837C137293760 @default.
- W3210368837 hasConceptScore W3210368837C147168706 @default.
- W3210368837 hasConceptScore W3210368837C151730666 @default.
- W3210368837 hasConceptScore W3210368837C154945302 @default.
- W3210368837 hasConceptScore W3210368837C177148314 @default.
- W3210368837 hasConceptScore W3210368837C186644900 @default.
- W3210368837 hasConceptScore W3210368837C202615002 @default.
- W3210368837 hasConceptScore W3210368837C204321447 @default.
- W3210368837 hasConceptScore W3210368837C206134035 @default.
- W3210368837 hasConceptScore W3210368837C2779343474 @default.
- W3210368837 hasConceptScore W3210368837C33923547 @default.
- W3210368837 hasConceptScore W3210368837C41008148 @default.
- W3210368837 hasConceptScore W3210368837C50644808 @default.
- W3210368837 hasConceptScore W3210368837C86803240 @default.
- W3210368837 hasLocation W32103688371 @default.
- W3210368837 hasOpenAccess W3210368837 @default.
- W3210368837 hasPrimaryLocation W32103688371 @default.
- W3210368837 hasRelatedWork W1581407678 @default.
- W3210368837 hasRelatedWork W1732222442 @default.
- W3210368837 hasRelatedWork W1793121960 @default.
- W3210368837 hasRelatedWork W2278108219 @default.
- W3210368837 hasRelatedWork W2291973609 @default.
- W3210368837 hasRelatedWork W2474920236 @default.
- W3210368837 hasRelatedWork W2623259071 @default.
- W3210368837 hasRelatedWork W2786698962 @default.
- W3210368837 hasRelatedWork W2787457957 @default.
- W3210368837 hasRelatedWork W2866343820 @default.
- W3210368837 hasRelatedWork W2897449544 @default.
- W3210368837 hasRelatedWork W2902538158 @default.
- W3210368837 hasRelatedWork W2904707853 @default.
- W3210368837 hasRelatedWork W2912481627 @default.
- W3210368837 hasRelatedWork W2952569483 @default.
- W3210368837 hasRelatedWork W2963753324 @default.
- W3210368837 hasRelatedWork W3012620775 @default.
- W3210368837 hasRelatedWork W3017057183 @default.
- W3210368837 hasRelatedWork W3131175409 @default.
- W3210368837 hasRelatedWork W3173043187 @default.
- W3210368837 isParatext "false" @default.
- W3210368837 isRetracted "false" @default.
- W3210368837 magId "3210368837" @default.
- W3210368837 workType "article" @default.