Enhancing Dynamic Hand Gesture Recognition through Optimized Feature Selection using Double Machine Learning

Causal machine learning combines causal inference and machine learning to understand and utilize causal relationships in data. While traditional machine learning focuses on missions of prediction and pattern recognition, causal machine learning goes a step further by revealing causal relationships between variables. In this research, we employ the double machine learning method to identify variables in the gesture recognition problem where independent variables have causal relationships with the final gesture. These variables are then selected for further classification and analysis. By comparing this approach with traditional feature selection methods, we find that the variables selected using double machine learning are more useful for classification and yield excellent results across different machine learning classification models. This new double machine learning based approach provides a valuable reference for researchers during the feature selection stage.