美国卡耐基梅隆大学Eric A. Yttri和德国波恩大学Martin K. Schwarz共同合作，近期取得重要工作进展。他们研究开发了A-SoiD工具，这是一个用于专业指导用户高效发现行为数据的主动学习平台。相关研究成果2024年2月21日在线发表于《自然—方法学》杂志上。

据介绍，为了识别和提取自然行为，有两种方法很流行：有监督和无监督。每种方法都有其自身的优势和劣势（例如，用户偏见、培训成本、复杂性和行动发现），用户在决策时必须考虑这些优势和劣势。

A-SOiD学习平台融合了这些优势，并克服了它们几个固有的缺点。A-SOiD使用普通训练数据的一小部分迭代学习用户定义的组，同时通过定向无监督分类实现扩展分类。在社交互动小鼠中，尽管需要的训练数据减少了85%，但A-SOiD的表现优于标准方法。

此外，它通过无监督分类分离了在行为学上不同的小鼠互动。研究人员使用非人灵长类动物和人类三维姿势数据观察到了类似的性能和效率。在这两种情况下，A-SOiD聚类定义的透明度通过博弈论方法揭示了监督分类的定义特征。为了便于使用，A-SOiD是一个直观的开源界面，可有效分割用户定义的行为和发现的子动作。

附：英文原文

Title: A-SOiD, an active-learning platform for expert-guided, data-efficient discovery of behavior

Author: Tillmann, Jens F., Hsu, Alexander I., Schwarz, Martin K., Yttri, Eric A.

Issue&Volume: 2024-02-21

Abstract: To identify and extract naturalistic behavior, two methods have become popular: supervised and unsupervised. Each approach carries its own strengths and weaknesses (for example, user bias, training cost, complexity and action discovery), which the user must consider in their decision. Here, an active-learning platform, A-SOiD, blends these strengths, and in doing so, overcomes several of their inherent drawbacks. A-SOiD iteratively learns user-defined groups with a fraction of the usual training data, while attaining expansive classification through directed unsupervised classification. In socially interacting mice, A-SOiD outperformed standard methods despite requiring 85% less training data. Additionally, it isolated ethologically distinct mouse interactions via unsupervised classification. We observed similar performance and efficiency using nonhuman primate and human three-dimensional pose data. In both cases, the transparency in A-SOiD’s cluster definitions revealed the defining features of the supervised classification through a game-theoretic approach. To facilitate use, A-SOiD comes as an intuitive, open-source interface for efficient segmentation of user-defined behaviors and discovered sub-actions.

DOI: 10.1038/s41592-024-02200-1

Source: https://www.nature.com/articles/s41592-024-02200-1