TY - GEN
T1 - A Comprehensive Neural and Behavioral Task Taxonomy Method for Transfer Learning in NLP
AU - Zhang, Yunhao
AU - Li, Chong
AU - Zhang, Xiaohan
AU - Dong, Xinyi
AU - Wang, Shaonan
N1 - Publisher Copyright:
© 2023 Asian Federation of Natural Language Processing.
PY - 2023/11
Y1 - 2023/11
N2 - Transfer learning is frequently utilized in scenarios with limited labeled examples, where a crucial step is to identify a related task to the target task. CogTaskonomy (Luo et al., 2022) was proposed to acquire a taxonomy of NLP tasks, specifically focusing on assessing the similarities between tasks. This method, inspired by cognitive processes, exhibits notable time efficiency. Nevertheless, it does not fully exploit the task-related information present in cognitive data and lacks a comprehensive evaluation of various types of cognitive data. To address these limitations, this paper proposes a comprehensive neural and behavioral method to investigate the relationship among NLP tasks. Our approach utilizes cognitive data, encompassing both neural data such as fMRI and EEG, as well as behavioral data including eye-tracking and semantic feature ratings. Each data modality is employed to establish a common representation space with Representation Similarity Analysis for projecting task-related representations. To fully leverage the cognitive information, we effectively extract the task-relevant information extracted from neural data through feature ranking. Experimental results on 12 NLP tasks demonstrate that our proposed method outperforms state-of-the-art methods on evaluating task similarity.
AB - Transfer learning is frequently utilized in scenarios with limited labeled examples, where a crucial step is to identify a related task to the target task. CogTaskonomy (Luo et al., 2022) was proposed to acquire a taxonomy of NLP tasks, specifically focusing on assessing the similarities between tasks. This method, inspired by cognitive processes, exhibits notable time efficiency. Nevertheless, it does not fully exploit the task-related information present in cognitive data and lacks a comprehensive evaluation of various types of cognitive data. To address these limitations, this paper proposes a comprehensive neural and behavioral method to investigate the relationship among NLP tasks. Our approach utilizes cognitive data, encompassing both neural data such as fMRI and EEG, as well as behavioral data including eye-tracking and semantic feature ratings. Each data modality is employed to establish a common representation space with Representation Similarity Analysis for projecting task-related representations. To fully leverage the cognitive information, we effectively extract the task-relevant information extracted from neural data through feature ranking. Experimental results on 12 NLP tasks demonstrate that our proposed method outperforms state-of-the-art methods on evaluating task similarity.
UR - https://www.scopus.com/pages/publications/85188539893
U2 - 10.18653/v1/2023.findings-ijcnlp.21
DO - 10.18653/v1/2023.findings-ijcnlp.21
M3 - Conference article published in proceeding or book
AN - SCOPUS:85188539893
T3 - IJCNLP-AACL 2023 - 13th International Joint Conference on Natural Language Processing and the 3rd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics, Findings of the Association for Computational Linguistics: IJCNLP-AACL 2023
SP - 233
EP - 241
BT - IJCNLP-AACL 2023 - 13th International JoinFindings of the Association for Computational Linguistics: IJCNLP-AACL 2023
A2 - Park, Jong C.
A2 - Arase, Yuki
A2 - Hu, Baotian
A2 - Lu, Wei
A2 - Wijaya, Derry
A2 - Purwarianti, Ayu
A2 - Krisnadhi, Adila Alfa
PB - Association for Computational Linguistics (ACL)
T2 - 13th International Joint Conference on Natural Language Processing and the 3rd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics: Findings of the Association for Computational Linguistic, IJCNLP-AACL 2023
Y2 - 1 November 2023 through 4 November 2023
ER -