Saved in:
Bibliographic Details
Main Authors: Kim, Doyoung, Yoon, Susik, Park, Dongmin, Lee, Youngjun, Song, Hwanjun, Bang, Jihwan, Lee, Jae-Gil
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2311.12048
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917728388907008
author Kim, Doyoung
Yoon, Susik
Park, Dongmin
Lee, Youngjun
Song, Hwanjun
Bang, Jihwan
Lee, Jae-Gil
author_facet Kim, Doyoung
Yoon, Susik
Park, Dongmin
Lee, Youngjun
Song, Hwanjun
Bang, Jihwan
Lee, Jae-Gil
contents In real-world continual learning (CL) scenarios, tasks often exhibit intricate and unpredictable semantic shifts, posing challenges for fixed prompt management strategies which are tailored to only handle semantic shifts of uniform degree (i.e., uniformly mild or uniformly abrupt). To address this limitation, we propose an adaptive prompting approach that effectively accommodates semantic shifts of varying degree where mild and abrupt shifts are mixed. AdaPromptCL employs the assign-and-refine semantic grouping mechanism that dynamically manages prompt groups in accordance with the semantic similarity between tasks, enhancing the quality of grouping through continuous refinement. Our experiment results demonstrate that AdaPromptCL outperforms existing prompting methods by up to 21.3%, especially in the benchmark datasets with diverse semantic shifts between tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2311_12048
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle One Size Fits All for Semantic Shifts: Adaptive Prompt Tuning for Continual Learning
Kim, Doyoung
Yoon, Susik
Park, Dongmin
Lee, Youngjun
Song, Hwanjun
Bang, Jihwan
Lee, Jae-Gil
Machine Learning
In real-world continual learning (CL) scenarios, tasks often exhibit intricate and unpredictable semantic shifts, posing challenges for fixed prompt management strategies which are tailored to only handle semantic shifts of uniform degree (i.e., uniformly mild or uniformly abrupt). To address this limitation, we propose an adaptive prompting approach that effectively accommodates semantic shifts of varying degree where mild and abrupt shifts are mixed. AdaPromptCL employs the assign-and-refine semantic grouping mechanism that dynamically manages prompt groups in accordance with the semantic similarity between tasks, enhancing the quality of grouping through continuous refinement. Our experiment results demonstrate that AdaPromptCL outperforms existing prompting methods by up to 21.3%, especially in the benchmark datasets with diverse semantic shifts between tasks.
title One Size Fits All for Semantic Shifts: Adaptive Prompt Tuning for Continual Learning
topic Machine Learning
url https://arxiv.org/abs/2311.12048