CauSkelNet: Causal Representation Learning for Human Behaviour Analysis

2024-09-23

arXiv.org

AI Summary

Weak model interpretability and the absence of joint dynamics and causal mechanism modeling hinder motion recognition. To address this, we propose the first two-stage skeletal causal discovery framework integrating the PC algorithm with KL divergence. Our method automatically identifies and quantifies causal relationships among skeletal joints, generating interpretable, robust, and scale-invariant skeletal representations; it further incorporates graph convolutional networks for causal-aware action modeling. Evaluated on the EmoPain dataset, our model achieves significant improvements in accuracy, F1-score, and recall—particularly enhancing discriminative capability for protective behaviors. Ablation studies demonstrate strong robustness to variations in training data scale. This work establishes a novel paradigm for human behavior analysis that jointly ensures causal interpretability and biomechanical plausibility.

Technology Category

• Knowledge Representation and Reasoning: Action, Change, and Causality
• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Constrained by the lack of model interpretability and a deep understanding of human movement in traditional movement recognition machine learning methods, this study introduces a novel representation learning method based on causal inference to better understand human joint dynamics and complex behaviors. We propose a two-stage framework that combines the Peter-Clark (PC) algorithm and Kullback-Leibler (KL) divergence to identify and quantify causal relationships between joints. Our method effectively captures interactions and produces interpretable, robust representations. Experiments on the EmoPain dataset show that our causal GCN outperforms traditional GCNs in accuracy, F1 score, and recall, especially in detecting protective behaviors. The model is also highly invariant to data scale changes, enhancing its reliability in practical applications. Our approach advances human motion analysis and paves the way for more adaptive intelligent healthcare solutions.

Problem

• Enhancing interpretability in human movement recognition models
• Quantifying causal relationships between human joints
• Improving accuracy in detecting protective behaviors

Innovation

• Causal inference for human joint relationships
• Two-stage PC algorithm and KL divergence
• Interpretable causal Graph Convolutional Network