Scholarlink

<p>Recent advances in artificial intelligence technology have brought increasing attention to data collection methods. When prior information is unavailable, space-filling designs offer a robust approach to initial data collection. Traditional space-filling designs distribute points uniformly across the entire design space but often fail to account for variations between subregions. To address this limitation, we propose a multi-stage data collection approach, called the uniform design based sequential design method, which iteratively adds a local uniform design for a subregion at each experimental stage. We also introduce a novel criterion for evaluating the importance of a subregion. Theoretical results and simulations validate the effectiveness of the proposed method, and an application to a high-dimensional image recognition task demonstrates its practical utility.</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

A multi-stage data collection strategy using uniform design based sequential designs

  • 期刊论文

  • 发表日期：

    2026年3月9日

查看全文
  • Li Zheng,
  • Min-Qian Liu,
  • Liuqing Yang

摘要

Recent advances in artificial intelligence technology have brought increasing attention to data collection methods. When prior information is unavailable, space-filling designs offer a robust approach to initial data collection. Traditional space-filling designs distribute points uniformly across the entire design space but often fail to account for variations between subregions. To address this limitation, we propose a multi-stage data collection approach, called the uniform design based sequential design method, which iteratively adds a local uniform design for a subregion at each experimental stage. We also introduce a novel criterion for evaluating the importance of a subregion. Theoretical results and simulations validate the effectiveness of the proposed method, and an application to a high-dimensional image recognition task demonstrates its practical utility.