Investigating How Infants Develop Awareness of Their Environment Using AI Analysis

Overview

A recent experiment delved into the emergence of purposeful action in infants by manipulating their connection to the environment.

Methodology

• Using Vicon motion capture data, Histograms of Joint Displacements (HJDs) were created to analyze 3D infant spatial trajectories.
• Various AI architectures including k-Nearest Neighbour and Convolutional Neural Networks were tested for accuracy in classifying movement data.
• Deep learning approaches, like 2D-CapsNet, showed higher accuracy in analyzing full-body features.

Key Findings

• Foot activity exhibited the most distinct pattern alterations, indicating a strong impact of interaction with the environment on infant behavior.
• Deep learning models, specifically 2D-CapsNet, displayed superior accuracy in classifying movement data.