Adaptive Knowledge Sharing in Multi-Task Learning: Insights from Electricity Data Analysis

Yu Hsiang Chang, Lo Pang Yun Ting, Wei Cheng Yin, Ko Wei Su, Kun Ta Chuang

研究成果: Conference contribution

摘要

In time-series machine learning, the challenge of obtaining labeled data has spurred interest in using unlabeled data for model training. Current research primarily focuses on deep multi-task learning, emphasizing the hard parameter-sharing approach. Importantly, when correlations between tasks are weak, indiscriminate parameter sharing can lead to learning interference. Consequently, we introduce a novel framework called DPS, which separates training into dependency-learning and parameter-sharing phases. This structure allows the model to manage knowledge sharing between tasks dynamically. Additionally, we introduce a loss function to align neuron functionalities across tasks, addressing learning interference. Through experiments on real-world datasets, we demonstrate the superiority of DPS over baselines. Moreover, our results shed light on the impacts of the two designed training phases, validating that DPS consistently ensures a degree of learning stability.

原文English
主出版物標題Trends and Applications in Knowledge Discovery and Data Mining - PAKDD 2024 Workshops, RAFDA and IWTA, Proceedings
編輯Zhaoxia Wang, Chang Wei Tan
發行者Springer Science and Business Media Deutschland GmbH
頁面148-160
頁數13
ISBN(列印)9789819726493
DOIs
出版狀態Published - 2024
事件Workshop on Research and Applications of Foundation Models for Data Mining and Affective Computing, RAFDA 2024 and International Workshop on Temporal Analytics, IWTA 2024 held in conjunction with the 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023 - Taipei, Taiwan
持續時間: 2024 5月 72024 5月 10

出版系列

名字Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
14658 LNAI
ISSN(列印)0302-9743
ISSN(電子)1611-3349

Conference

ConferenceWorkshop on Research and Applications of Foundation Models for Data Mining and Affective Computing, RAFDA 2024 and International Workshop on Temporal Analytics, IWTA 2024 held in conjunction with the 27th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2023
國家/地區Taiwan
城市Taipei
期間24-05-0724-05-10

All Science Journal Classification (ASJC) codes

  • 理論電腦科學
  • 一般電腦科學

指紋

深入研究「Adaptive Knowledge Sharing in Multi-Task Learning: Insights from Electricity Data Analysis」主題。共同形成了獨特的指紋。

引用此