Matches in SemOpenAlex for { <https://semopenalex.org/work/W3043669249> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W3043669249 abstract "We focus on the problem of how to achieve online continual learning under memory-constrained conditions where the input data may not be known a priori. These constraints are relevant in edge computing scenarios. We have developed an architecture where input processing over data streams and online learning are integrated in a single recurrent network architecture. This allows us to cast metalearning optimization as a mixed-integer optimization problem, where different synaptic plasticity algorithms and feature extraction layers can be swapped out and their hyperparameters are optimized to identify optimal architectures for different sets of tasks. We utilize a Bayesian optimization method to search over a design space that spans multiple learning algorithms, their specific hyperparameters, and feature extraction layers. We demonstrate our approach for online non-incremental and class-incremental learning tasks. Our optimization algorithm finds configurations that achieve superior continual learning performance on Split-MNIST and Permuted-MNIST data as compared with other memory-constrained learning approaches, and it matches that of the state-of-the-art memory replay-based approaches without explicit data storage and replay. Our approach allows us to explore the transferability of optimal learning conditions to tasks and datasets that have not been previously seen. We demonstrate that the accuracy of our transfer metalearning across datasets can be largely explained through a transfer coefficient that can be based on metrics of dimensionality and distance between datasets." @default.
- W3043669249 created "2020-07-23" @default.
- W3043669249 creator A5004608167 @default.
- W3043669249 creator A5027532222 @default.
- W3043669249 creator A5068721920 @default.
- W3043669249 date "2020-06-12" @default.
- W3043669249 modified "2023-10-02" @default.
- W3043669249 title "Multilayer Neuromodulated Architectures for Memory-Constrained Online Continual Learning" @default.
- W3043669249 cites W1512746852 @default.
- W3043669249 cites W2113839990 @default.
- W3043669249 cites W2170413022 @default.
- W3043669249 cites W2401231614 @default.
- W3043669249 cites W2588048611 @default.
- W3043669249 cites W2618767506 @default.
- W3043669249 cites W2734314755 @default.
- W3043669249 cites W2734358244 @default.
- W3043669249 cites W2737492962 @default.
- W3043669249 cites W2750384547 @default.
- W3043669249 cites W2765101016 @default.
- W3043669249 cites W2788388592 @default.
- W3043669249 cites W2804175194 @default.
- W3043669249 cites W2804746922 @default.
- W3043669249 cites W2894094671 @default.
- W3043669249 cites W2899063268 @default.
- W3043669249 cites W2913525628 @default.
- W3043669249 cites W2952571470 @default.
- W3043669249 cites W2952677972 @default.
- W3043669249 cites W2963168530 @default.
- W3043669249 cites W2963588172 @default.
- W3043669249 cites W2966927517 @default.
- W3043669249 cites W2967925425 @default.
- W3043669249 cites W2970505118 @default.
- W3043669249 cites W2971983254 @default.
- W3043669249 cites W2978416425 @default.
- W3043669249 cites W2997946860 @default.
- W3043669249 cites W3008449794 @default.
- W3043669249 cites W3013325675 @default.
- W3043669249 cites W3013343111 @default.
- W3043669249 cites W3017160304 @default.
- W3043669249 cites W3035170382 @default.
- W3043669249 cites W3098061682 @default.
- W3043669249 cites W3125116114 @default.
- W3043669249 cites W99485931 @default.
- W3043669249 cites W2426267443 @default.
- W3043669249 hasPublicationYear "2020" @default.
- W3043669249 type Work @default.
- W3043669249 sameAs 3043669249 @default.
- W3043669249 citedByCount "0" @default.
- W3043669249 crossrefType "posted-content" @default.
- W3043669249 hasAuthorship W3043669249A5004608167 @default.
- W3043669249 hasAuthorship W3043669249A5027532222 @default.
- W3043669249 hasAuthorship W3043669249A5068721920 @default.
- W3043669249 hasConcept C108583219 @default.
- W3043669249 hasConcept C111030470 @default.
- W3043669249 hasConcept C119857082 @default.
- W3043669249 hasConcept C150899416 @default.
- W3043669249 hasConcept C154945302 @default.
- W3043669249 hasConcept C190502265 @default.
- W3043669249 hasConcept C2778049539 @default.
- W3043669249 hasConcept C41008148 @default.
- W3043669249 hasConcept C8642999 @default.
- W3043669249 hasConceptScore W3043669249C108583219 @default.
- W3043669249 hasConceptScore W3043669249C111030470 @default.
- W3043669249 hasConceptScore W3043669249C119857082 @default.
- W3043669249 hasConceptScore W3043669249C150899416 @default.
- W3043669249 hasConceptScore W3043669249C154945302 @default.
- W3043669249 hasConceptScore W3043669249C190502265 @default.
- W3043669249 hasConceptScore W3043669249C2778049539 @default.
- W3043669249 hasConceptScore W3043669249C41008148 @default.
- W3043669249 hasConceptScore W3043669249C8642999 @default.
- W3043669249 hasLocation W30436692491 @default.
- W3043669249 hasOpenAccess W3043669249 @default.
- W3043669249 hasPrimaryLocation W30436692491 @default.
- W3043669249 hasRelatedWork W2039044448 @default.
- W3043669249 hasRelatedWork W2604763608 @default.
- W3043669249 hasRelatedWork W2607462057 @default.
- W3043669249 hasRelatedWork W2610817424 @default.
- W3043669249 hasRelatedWork W2750881831 @default.
- W3043669249 hasRelatedWork W2929907253 @default.
- W3043669249 hasRelatedWork W2976371263 @default.
- W3043669249 hasRelatedWork W2977426525 @default.
- W3043669249 hasRelatedWork W2978291687 @default.
- W3043669249 hasRelatedWork W2981966198 @default.
- W3043669249 hasRelatedWork W2982659598 @default.
- W3043669249 hasRelatedWork W2986368801 @default.
- W3043669249 hasRelatedWork W2993852105 @default.
- W3043669249 hasRelatedWork W3013609115 @default.
- W3043669249 hasRelatedWork W3034837814 @default.
- W3043669249 hasRelatedWork W3037841817 @default.
- W3043669249 hasRelatedWork W3127888919 @default.
- W3043669249 hasRelatedWork W3165418502 @default.
- W3043669249 hasRelatedWork W3180392831 @default.
- W3043669249 hasRelatedWork W995784193 @default.
- W3043669249 isParatext "false" @default.
- W3043669249 isRetracted "false" @default.
- W3043669249 magId "3043669249" @default.
- W3043669249 workType "article" @default.