Scholarlink

<p>Unsupervised Domain Adaptation (UDA) aims to harness labeled source data to train models for unlabeled target data. While research in UDA is extensive in domains like computer vision and natural language processing, it remains under-explored for time series data despite its widespread real-world applications. Our paper addresses this gap by introducing a comprehensive benchmark for evaluating UDA techniques for time series classification, with a focus on deep learning methods. We provide a fair and standardized UDA method assessment with state of the art neural network backbones (e.g. Inception) for time series data, as well as seven extra datasets in addition to the already established ones. This benchmark offers insights into the strengths and limitations of the evaluated approaches while preserving the unsupervised nature of DA, making it directly applicable to real data mining problems. It serves as a beneficial resource for researchers and practitioners, fostering innovation in this critical field. In order to ensure reproducibility, the code and results have been open-sourced.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deep unsupervised domain adaptation for time series classification: a benchmark

  • 期刊论文

  • 发表日期：

    2025年6月9日

查看全文
  • Hassan Ismail Fawaz,
  • Ganesh Del Grosso,
  • Tanguy Kerdoncuff,
  • Aurélie Boisbunon,
  • Illyyne Saffar

摘要

Unsupervised Domain Adaptation (UDA) aims to harness labeled source data to train models for unlabeled target data. While research in UDA is extensive in domains like computer vision and natural language processing, it remains under-explored for time series data despite its widespread real-world applications. Our paper addresses this gap by introducing a comprehensive benchmark for evaluating UDA techniques for time series classification, with a focus on deep learning methods. We provide a fair and standardized UDA method assessment with state of the art neural network backbones (e.g. Inception) for time series data, as well as seven extra datasets in addition to the already established ones. This benchmark offers insights into the strengths and limitations of the evaluated approaches while preserving the unsupervised nature of DA, making it directly applicable to real data mining problems. It serves as a beneficial resource for researchers and practitioners, fostering innovation in this critical field. In order to ensure reproducibility, the code and results have been open-sourced.