Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385570843> ?p ?o ?g. }
Showing items 1 to 59 of
59
with 100 items per page.
- W4385570843 abstract "State-of-the-art few-shot learning (FSL) methods leverage prompt-based fine-tuning to obtain remarkable results for natural language understanding (NLU) tasks. While much of the prior FSL methods focus on improving downstream task performance, there is a limited understanding of the adversarial robustness of such methods. In this work, we conduct an extensive study of several state-of-the-art FSL methods to assess their robustness to adversarial perturbations. To better understand the impact of various factors towards robustness (or the lack of it), we evaluate prompt-based FSL methods against fully fine-tuned models for aspects such as the use of unlabeled data, multiple prompts, number of few-shot examples, model size and type. Our results on six GLUE tasks indicate that compared to fully fine-tuned models, vanilla FSL methods lead to a notable relative drop in task performance (i.e., are less robust) in the face of adversarial perturbations. However, using (i) unlabeled data for prompt-based FSL and (ii) multiple prompts flip the trend – the few-shot learning approaches demonstrate a lesser drop in task performance than fully fine-tuned models. We further demonstrate that increasing the number of few-shot examples and model size lead to increased adversarial robustness of vanilla FSL methods. Broadly, our work sheds light on the adversarial robustness evaluation of prompt-based FSL methods for NLU tasks." @default.
- W4385570843 created "2023-08-05" @default.
- W4385570843 creator A5033994052 @default.
- W4385570843 creator A5056142478 @default.
- W4385570843 creator A5058032023 @default.
- W4385570843 creator A5092240627 @default.
- W4385570843 date "2023-01-01" @default.
- W4385570843 modified "2023-09-27" @default.
- W4385570843 title "Adversarial Robustness of Prompt-based Few-Shot Learning for Natural Language Understanding" @default.
- W4385570843 doi "https://doi.org/10.18653/v1/2023.findings-acl.138" @default.
- W4385570843 hasPublicationYear "2023" @default.
- W4385570843 type Work @default.
- W4385570843 citedByCount "0" @default.
- W4385570843 crossrefType "proceedings-article" @default.
- W4385570843 hasAuthorship W4385570843A5033994052 @default.
- W4385570843 hasAuthorship W4385570843A5056142478 @default.
- W4385570843 hasAuthorship W4385570843A5058032023 @default.
- W4385570843 hasAuthorship W4385570843A5092240627 @default.
- W4385570843 hasBestOaLocation W43855708431 @default.
- W4385570843 hasConcept C104317684 @default.
- W4385570843 hasConcept C108583219 @default.
- W4385570843 hasConcept C119857082 @default.
- W4385570843 hasConcept C153083717 @default.
- W4385570843 hasConcept C154945302 @default.
- W4385570843 hasConcept C185592680 @default.
- W4385570843 hasConcept C195324797 @default.
- W4385570843 hasConcept C2779439875 @default.
- W4385570843 hasConcept C37736160 @default.
- W4385570843 hasConcept C41008148 @default.
- W4385570843 hasConcept C55493867 @default.
- W4385570843 hasConcept C63479239 @default.
- W4385570843 hasConceptScore W4385570843C104317684 @default.
- W4385570843 hasConceptScore W4385570843C108583219 @default.
- W4385570843 hasConceptScore W4385570843C119857082 @default.
- W4385570843 hasConceptScore W4385570843C153083717 @default.
- W4385570843 hasConceptScore W4385570843C154945302 @default.
- W4385570843 hasConceptScore W4385570843C185592680 @default.
- W4385570843 hasConceptScore W4385570843C195324797 @default.
- W4385570843 hasConceptScore W4385570843C2779439875 @default.
- W4385570843 hasConceptScore W4385570843C37736160 @default.
- W4385570843 hasConceptScore W4385570843C41008148 @default.
- W4385570843 hasConceptScore W4385570843C55493867 @default.
- W4385570843 hasConceptScore W4385570843C63479239 @default.
- W4385570843 hasLocation W43855708431 @default.
- W4385570843 hasOpenAccess W4385570843 @default.
- W4385570843 hasPrimaryLocation W43855708431 @default.
- W4385570843 hasRelatedWork W3193857078 @default.
- W4385570843 hasRelatedWork W3208304128 @default.
- W4385570843 hasRelatedWork W4221150850 @default.
- W4385570843 hasRelatedWork W4223943233 @default.
- W4385570843 hasRelatedWork W4224952001 @default.
- W4385570843 hasRelatedWork W4312200629 @default.
- W4385570843 hasRelatedWork W4360585206 @default.
- W4385570843 hasRelatedWork W4364306694 @default.
- W4385570843 hasRelatedWork W4380075502 @default.
- W4385570843 hasRelatedWork W4382463655 @default.
- W4385570843 isParatext "false" @default.
- W4385570843 isRetracted "false" @default.
- W4385570843 workType "article" @default.