Yifan Tang, Mostafa Rahmani Dehaghani, Pouyan Sajadi, G. Gary Wang
{"title":"为转移学习选择源数据子集,并将其应用于金属增材制造","authors":"Yifan Tang, Mostafa Rahmani Dehaghani, Pouyan Sajadi, G. Gary Wang","doi":"10.1007/s10845-024-02402-6","DOIUrl":null,"url":null,"abstract":"<p>Considering data insufficiency in metal additive manufacturing (AM), transfer learning (TL) has been adopted to extract knowledge from source domains (e.g., completed printings) to improve the modeling performance in target domains (e.g., new printings). Current applications use all accessible source data directly in TL with no regard to the similarity between source and target data. This paper proposes a systematic method to find appropriate subsets of source data based on similarities between the source and limited target datasets. Such similarity is characterized by the spatial and model distance metrics. A Pareto frontier-based source data selection method is developed, where the source data located on the Pareto frontier defined by two similarity distance metrics are selected iteratively. This method is integrated into an instance-based TL method (decision tree regression model) and a model-based TL method (fine-tuned artificial neural network). Both models are then tested on several regression tasks in metal AM. Comparison results demonstrate that (1) the source data selection method is general and supports integration with various TL methods and distance metrics, (2) compared with using all source data, the proposed method can find a subset of source data from the same domain with better TL performance in metal AM regression tasks involving different processes and machines, and (3) when multiple source domains exist, the source data selection method could find the subset from one source domain to obtain comparable or better TL performance than the model constructed using data from all source domains.</p>","PeriodicalId":16193,"journal":{"name":"Journal of Intelligent Manufacturing","volume":"3 1","pages":""},"PeriodicalIF":5.9000,"publicationDate":"2024-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Selecting subsets of source data for transfer learning with applications in metal additive manufacturing\",\"authors\":\"Yifan Tang, Mostafa Rahmani Dehaghani, Pouyan Sajadi, G. Gary Wang\",\"doi\":\"10.1007/s10845-024-02402-6\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>Considering data insufficiency in metal additive manufacturing (AM), transfer learning (TL) has been adopted to extract knowledge from source domains (e.g., completed printings) to improve the modeling performance in target domains (e.g., new printings). Current applications use all accessible source data directly in TL with no regard to the similarity between source and target data. This paper proposes a systematic method to find appropriate subsets of source data based on similarities between the source and limited target datasets. Such similarity is characterized by the spatial and model distance metrics. A Pareto frontier-based source data selection method is developed, where the source data located on the Pareto frontier defined by two similarity distance metrics are selected iteratively. This method is integrated into an instance-based TL method (decision tree regression model) and a model-based TL method (fine-tuned artificial neural network). Both models are then tested on several regression tasks in metal AM. Comparison results demonstrate that (1) the source data selection method is general and supports integration with various TL methods and distance metrics, (2) compared with using all source data, the proposed method can find a subset of source data from the same domain with better TL performance in metal AM regression tasks involving different processes and machines, and (3) when multiple source domains exist, the source data selection method could find the subset from one source domain to obtain comparable or better TL performance than the model constructed using data from all source domains.</p>\",\"PeriodicalId\":16193,\"journal\":{\"name\":\"Journal of Intelligent Manufacturing\",\"volume\":\"3 1\",\"pages\":\"\"},\"PeriodicalIF\":5.9000,\"publicationDate\":\"2024-05-12\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Intelligent Manufacturing\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://doi.org/10.1007/s10845-024-02402-6\",\"RegionNum\":2,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Intelligent Manufacturing","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.1007/s10845-024-02402-6","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}
引用次数: 0
摘要
考虑到金属增材制造(AM)中的数据不足,人们采用了迁移学习(TL)技术,从源域(如已完成的印刷)中提取知识,以提高目标域(如新的印刷)的建模性能。目前的应用是在 TL 中直接使用所有可访问的源数据,而不考虑源数据和目标数据之间的相似性。本文提出了一种系统方法,可根据源数据集和有限目标数据集之间的相似性找到适当的源数据子集。这种相似性由空间距离和模型距离度量表征。本文开发了一种基于帕累托前沿的源数据选择方法,通过迭代选择位于由两个相似性距离指标定义的帕累托前沿上的源数据。这种方法被集成到基于实例的 TL 方法(决策树回归模型)和基于模型的 TL 方法(微调人工神经网络)中。然后,在金属 AM 的若干回归任务中对这两种模型进行了测试。比较结果表明:(1) 源数据选择方法具有通用性,支持与各种 TL 方法和距离度量的集成;(2) 与使用所有源数据相比,在涉及不同流程和机器的金属 AM 回归任务中,所提出的方法可以从同一域中找到具有更好 TL 性能的源数据子集;(3) 当存在多个源域时,源数据选择方法可以从一个源域中找到子集,从而获得与使用所有源域数据构建的模型相当或更好的 TL 性能。
Selecting subsets of source data for transfer learning with applications in metal additive manufacturing
Considering data insufficiency in metal additive manufacturing (AM), transfer learning (TL) has been adopted to extract knowledge from source domains (e.g., completed printings) to improve the modeling performance in target domains (e.g., new printings). Current applications use all accessible source data directly in TL with no regard to the similarity between source and target data. This paper proposes a systematic method to find appropriate subsets of source data based on similarities between the source and limited target datasets. Such similarity is characterized by the spatial and model distance metrics. A Pareto frontier-based source data selection method is developed, where the source data located on the Pareto frontier defined by two similarity distance metrics are selected iteratively. This method is integrated into an instance-based TL method (decision tree regression model) and a model-based TL method (fine-tuned artificial neural network). Both models are then tested on several regression tasks in metal AM. Comparison results demonstrate that (1) the source data selection method is general and supports integration with various TL methods and distance metrics, (2) compared with using all source data, the proposed method can find a subset of source data from the same domain with better TL performance in metal AM regression tasks involving different processes and machines, and (3) when multiple source domains exist, the source data selection method could find the subset from one source domain to obtain comparable or better TL performance than the model constructed using data from all source domains.
期刊介绍:
The Journal of Nonlinear Engineering aims to be a platform for sharing original research results in theoretical, experimental, practical, and applied nonlinear phenomena within engineering. It serves as a forum to exchange ideas and applications of nonlinear problems across various engineering disciplines. Articles are considered for publication if they explore nonlinearities in engineering systems, offering realistic mathematical modeling, utilizing nonlinearity for new designs, stabilizing systems, understanding system behavior through nonlinearity, optimizing systems based on nonlinear interactions, and developing algorithms to harness and leverage nonlinear elements.