Autoencoders effectively score athlete performance from wearable data

This paper investigates the effectiveness of various autoencoder architectures in compressing complex wearable telemetry data from runners into a single performance score. Researchers evaluated five dimensionality reduction models, including three autoencoder variants, PCA, and a Variational Autoencoder, assessing them on reconstruction error and the interpretability of the resulting latent score. The study found that a deep autoencoder performed best in terms of both low reconstruction error and a high composite score, indicating its suitability for athlete performance analysis. AI

IMPACT This research demonstrates a method for extracting meaningful performance metrics from raw wearable data, potentially improving athlete training and analysis.

RANK_REASON The cluster contains a research paper detailing new model architectures and their application. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Autoencoder
• PCA
• Variational Autoencoder

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →