Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313524871> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W4313524871 abstract "Recent years have seen a growing trend of deploying deep neural network-based applications on edge devices. Many of these applications, such as biometric identification, activity tracking, user preference learning, etc., require fine-tuning of the trained networks for user personalization. One way to prepare these models to handle new, unseen tasks, is to pre-train them on a distribution of known tasks. This observation has led to increasing research into meta-learning based few-shot learning techniques. However, basic meta-learning approaches do not account for the limited memory and computational resources during on-chip training. We propose a modified meta-learning algorithm that enables quantized fine-tuning to optimally condition the models for on-chip few shot learning. The modification involves the inclusion of target hardware constraints upfront in the meta-learning process. Block floating point datatypes with low precision mantissa bits are utilized in the forward and backward passes, to allow hardware-friendly adaptation. Experiments show that our algorithm provides better initializations than conventional algorithms, more suitable for efficient quantized fine-tuning. This allows the few-shot learner to achieve better convergence, in terms of accuracy and speed. Extensive experiments are also performed to analyze the impact of initialization on quantized fine-tuning and further corroborate the benefits of our method." @default.
- W4313524871 created "2023-01-06" @default.
- W4313524871 creator A5012150553 @default.
- W4313524871 creator A5030379454 @default.
- W4313524871 creator A5033235745 @default.
- W4313524871 creator A5059602336 @default.
- W4313524871 creator A5062896285 @default.
- W4313524871 creator A5082798974 @default.
- W4313524871 date "2022-12-01" @default.
- W4313524871 modified "2023-10-03" @default.
- W4313524871 title "Learn to Learn on Chip: Hardware-aware Meta-learning for Quantized Few-shot Learning at the Edge" @default.
- W4313524871 cites W2194321275 @default.
- W4313524871 cites W2913318911 @default.
- W4313524871 cites W2944138435 @default.
- W4313524871 cites W2945854677 @default.
- W4313524871 cites W2946337031 @default.
- W4313524871 cites W2963679555 @default.
- W4313524871 cites W2983307807 @default.
- W4313524871 cites W2988640543 @default.
- W4313524871 cites W3015764133 @default.
- W4313524871 cites W3034942609 @default.
- W4313524871 cites W3035964421 @default.
- W4313524871 cites W3047691619 @default.
- W4313524871 cites W3091977376 @default.
- W4313524871 cites W3176714321 @default.
- W4313524871 cites W3183680883 @default.
- W4313524871 cites W3195577076 @default.
- W4313524871 doi "https://doi.org/10.1109/sec54971.2022.00009" @default.
- W4313524871 hasPublicationYear "2022" @default.
- W4313524871 type Work @default.
- W4313524871 citedByCount "0" @default.
- W4313524871 crossrefType "proceedings-article" @default.
- W4313524871 hasAuthorship W4313524871A5012150553 @default.
- W4313524871 hasAuthorship W4313524871A5030379454 @default.
- W4313524871 hasAuthorship W4313524871A5033235745 @default.
- W4313524871 hasAuthorship W4313524871A5059602336 @default.
- W4313524871 hasAuthorship W4313524871A5062896285 @default.
- W4313524871 hasAuthorship W4313524871A5082798974 @default.
- W4313524871 hasConcept C108583219 @default.
- W4313524871 hasConcept C111919701 @default.
- W4313524871 hasConcept C113775141 @default.
- W4313524871 hasConcept C114466953 @default.
- W4313524871 hasConcept C119857082 @default.
- W4313524871 hasConcept C138236772 @default.
- W4313524871 hasConcept C154945302 @default.
- W4313524871 hasConcept C162307627 @default.
- W4313524871 hasConcept C162324750 @default.
- W4313524871 hasConcept C187736073 @default.
- W4313524871 hasConcept C199360897 @default.
- W4313524871 hasConcept C2780451532 @default.
- W4313524871 hasConcept C2781002164 @default.
- W4313524871 hasConcept C41008148 @default.
- W4313524871 hasConcept C79974875 @default.
- W4313524871 hasConcept C98045186 @default.
- W4313524871 hasConceptScore W4313524871C108583219 @default.
- W4313524871 hasConceptScore W4313524871C111919701 @default.
- W4313524871 hasConceptScore W4313524871C113775141 @default.
- W4313524871 hasConceptScore W4313524871C114466953 @default.
- W4313524871 hasConceptScore W4313524871C119857082 @default.
- W4313524871 hasConceptScore W4313524871C138236772 @default.
- W4313524871 hasConceptScore W4313524871C154945302 @default.
- W4313524871 hasConceptScore W4313524871C162307627 @default.
- W4313524871 hasConceptScore W4313524871C162324750 @default.
- W4313524871 hasConceptScore W4313524871C187736073 @default.
- W4313524871 hasConceptScore W4313524871C199360897 @default.
- W4313524871 hasConceptScore W4313524871C2780451532 @default.
- W4313524871 hasConceptScore W4313524871C2781002164 @default.
- W4313524871 hasConceptScore W4313524871C41008148 @default.
- W4313524871 hasConceptScore W4313524871C79974875 @default.
- W4313524871 hasConceptScore W4313524871C98045186 @default.
- W4313524871 hasFunder F4320327336 @default.
- W4313524871 hasLocation W43135248711 @default.
- W4313524871 hasOpenAccess W4313524871 @default.
- W4313524871 hasPrimaryLocation W43135248711 @default.
- W4313524871 hasRelatedWork W3014300295 @default.
- W4313524871 hasRelatedWork W3164822677 @default.
- W4313524871 hasRelatedWork W3177307486 @default.
- W4313524871 hasRelatedWork W3200361725 @default.
- W4313524871 hasRelatedWork W4223943233 @default.
- W4313524871 hasRelatedWork W4225161397 @default.
- W4313524871 hasRelatedWork W4309045103 @default.
- W4313524871 hasRelatedWork W4312200629 @default.
- W4313524871 hasRelatedWork W4360585206 @default.
- W4313524871 hasRelatedWork W4364306694 @default.
- W4313524871 isParatext "false" @default.
- W4313524871 isRetracted "false" @default.
- W4313524871 workType "article" @default.