Cross-domain topic transfer learning method based on multiple balance and feature fusion

In transfer learning, traditional homogeneous transfer learning assumes similar data and feature distributions between the source and target domains, focusing primarily on parameter sharing to enhance model performance. However, heterogeneous transfer learning for topic model, disparities in data an...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhenshun Xu, Zhenbiao Wang, Wenhao Zhang, Zengjin Tang
Format: Article
Language:English
Published: Elsevier 2025-05-01
Series:Heliyon
Online Access:http://www.sciencedirect.com/science/article/pii/S2405844024167941
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849694923020828672
author Zhenshun Xu
Zhenbiao Wang
Wenhao Zhang
Zengjin Tang
author_facet Zhenshun Xu
Zhenbiao Wang
Wenhao Zhang
Zengjin Tang
author_sort Zhenshun Xu
collection DOAJ
description In transfer learning, traditional homogeneous transfer learning assumes similar data and feature distributions between the source and target domains, focusing primarily on parameter sharing to enhance model performance. However, heterogeneous transfer learning for topic model, disparities in data and feature distributions lead to negative transfer, diminishing the effectiveness of topic extraction. This research explores alternative methods for heterogeneous transfer learning in topic model beyond traditional parameter sharing, seeking to maximize the utilization of source domain knowledge and features, mitigating the interference of negative transfer. We propose a novel approach for cross-domain topic transfer learning by combining feature fusion and balancing data and labels. Including applying dual-supervised techniques for handling label dependencies, employing function constraints and data enhancement to adjust data distributions, and utilizing feature fusion techniques to mitigate feature distribution disparities. Additionally, we introduce a topic knowledge distillation method, leveraging topics from the source domain to guide and optimize topic generation in the target domain. In practical applications, this method enhances animal disease topic mining by integrating feature knowledge from the animal diseases dataset and the 20 Newsgroups dataset. The experiments verify the effectiveness of this method, bridging gaps in topic generation and advancing the application of topic modeling in complex and dynamic contexts.
format Article
id doaj-art-88dbe391ef17442ea8e17496ed3bc664
institution DOAJ
issn 2405-8440
language English
publishDate 2025-05-01
publisher Elsevier
record_format Article
series Heliyon
spelling doaj-art-88dbe391ef17442ea8e17496ed3bc6642025-08-20T03:19:56ZengElsevierHeliyon2405-84402025-05-011110e4076310.1016/j.heliyon.2024.e40763Cross-domain topic transfer learning method based on multiple balance and feature fusionZhenshun Xu0Zhenbiao Wang1Wenhao Zhang2Zengjin Tang3College of Compute Science and Engineering, North Minzu University, Yinchuan, China; The Key Laboratory of Images & Graphics Intelligent Processing of State Ethnic Affairs Commission, North Minzu University, Yinchuan, China; Corresponding author. College of Compute Science and Engineering, North Minzu University, Yinchuan, China.College of Compute Science and Engineering, North Minzu University, Yinchuan, China; The Key Laboratory of Images & Graphics Intelligent Processing of State Ethnic Affairs Commission, North Minzu University, Yinchuan, ChinaCollege of Compute Science and Engineering, North Minzu University, Yinchuan, China; The Key Laboratory of Images & Graphics Intelligent Processing of State Ethnic Affairs Commission, North Minzu University, Yinchuan, ChinaCollege of Compute Science and Engineering, North Minzu University, Yinchuan, China; The Key Laboratory of Images & Graphics Intelligent Processing of State Ethnic Affairs Commission, North Minzu University, Yinchuan, ChinaIn transfer learning, traditional homogeneous transfer learning assumes similar data and feature distributions between the source and target domains, focusing primarily on parameter sharing to enhance model performance. However, heterogeneous transfer learning for topic model, disparities in data and feature distributions lead to negative transfer, diminishing the effectiveness of topic extraction. This research explores alternative methods for heterogeneous transfer learning in topic model beyond traditional parameter sharing, seeking to maximize the utilization of source domain knowledge and features, mitigating the interference of negative transfer. We propose a novel approach for cross-domain topic transfer learning by combining feature fusion and balancing data and labels. Including applying dual-supervised techniques for handling label dependencies, employing function constraints and data enhancement to adjust data distributions, and utilizing feature fusion techniques to mitigate feature distribution disparities. Additionally, we introduce a topic knowledge distillation method, leveraging topics from the source domain to guide and optimize topic generation in the target domain. In practical applications, this method enhances animal disease topic mining by integrating feature knowledge from the animal diseases dataset and the 20 Newsgroups dataset. The experiments verify the effectiveness of this method, bridging gaps in topic generation and advancing the application of topic modeling in complex and dynamic contexts.http://www.sciencedirect.com/science/article/pii/S2405844024167941
spellingShingle Zhenshun Xu
Zhenbiao Wang
Wenhao Zhang
Zengjin Tang
Cross-domain topic transfer learning method based on multiple balance and feature fusion
Heliyon
title Cross-domain topic transfer learning method based on multiple balance and feature fusion
title_full Cross-domain topic transfer learning method based on multiple balance and feature fusion
title_fullStr Cross-domain topic transfer learning method based on multiple balance and feature fusion
title_full_unstemmed Cross-domain topic transfer learning method based on multiple balance and feature fusion
title_short Cross-domain topic transfer learning method based on multiple balance and feature fusion
title_sort cross domain topic transfer learning method based on multiple balance and feature fusion
url http://www.sciencedirect.com/science/article/pii/S2405844024167941
work_keys_str_mv AT zhenshunxu crossdomaintopictransferlearningmethodbasedonmultiplebalanceandfeaturefusion
AT zhenbiaowang crossdomaintopictransferlearningmethodbasedonmultiplebalanceandfeaturefusion
AT wenhaozhang crossdomaintopictransferlearningmethodbasedonmultiplebalanceandfeaturefusion
AT zengjintang crossdomaintopictransferlearningmethodbasedonmultiplebalanceandfeaturefusion