CauSkelNet: Causal Representation Learning for Human Behaviour Analysis

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gu, Xingrui, Jiang, Chuyi, Wang, Erte, Cui, Qiang, Tian, Leimin, Wu, Lianlong, Song, Siyang, Yu, Chuang
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918075484340224
author Gu, Xingrui
Jiang, Chuyi
Wang, Erte
Cui, Qiang
Tian, Leimin
Wu, Lianlong
Song, Siyang
Yu, Chuang
author_facet Gu, Xingrui
Jiang, Chuyi
Wang, Erte
Cui, Qiang
Tian, Leimin
Wu, Lianlong
Song, Siyang
Yu, Chuang
contents Traditional machine learning methods for movement recognition often struggle with limited model interpretability and a lack of insight into human movement dynamics. This study introduces a novel representation learning framework based on causal inference to address these challenges. Our two-stage approach combines the Peter-Clark (PC) algorithm and Kullback-Leibler (KL) divergence to identify and quantify causal relationships between human joints. By capturing joint interactions, the proposed causal Graph Convolutional Network (GCN) produces interpretable and robust representations. Experimental results on the EmoPain dataset demonstrate that the causal GCN outperforms traditional GCNs in accuracy, F1 score, and recall, particularly in detecting protective behaviors. This work contributes to advancing human motion analysis and lays a foundation for adaptive and intelligent healthcare solutions.
format Preprint
id arxiv_https___arxiv_org_abs_2409_15564
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle CauSkelNet: Causal Representation Learning for Human Behaviour Analysis
Gu, Xingrui
Jiang, Chuyi
Wang, Erte
Cui, Qiang
Tian, Leimin
Wu, Lianlong
Song, Siyang
Yu, Chuang
Machine Learning
Computer Vision and Pattern Recognition
Traditional machine learning methods for movement recognition often struggle with limited model interpretability and a lack of insight into human movement dynamics. This study introduces a novel representation learning framework based on causal inference to address these challenges. Our two-stage approach combines the Peter-Clark (PC) algorithm and Kullback-Leibler (KL) divergence to identify and quantify causal relationships between human joints. By capturing joint interactions, the proposed causal Graph Convolutional Network (GCN) produces interpretable and robust representations. Experimental results on the EmoPain dataset demonstrate that the causal GCN outperforms traditional GCNs in accuracy, F1 score, and recall, particularly in detecting protective behaviors. This work contributes to advancing human motion analysis and lays a foundation for adaptive and intelligent healthcare solutions.
title CauSkelNet: Causal Representation Learning for Human Behaviour Analysis
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2409.15564