UC San Diego

Predicting Quality of Video Gaming Experience Using Global-Scale Telemetry Data and Federated Learning

Predicting video game performance through hardware requirements often fails to account for the complex interplay between diverse user configurations and software optimization.

• A federated learning model using unique 'learnable knowledge kernels' (LKK) for players and games can accurately predict FPS performance while maintaining user privacy by avoiding centralized data collection.
• The model achieves high predictive accuracy by combining hardware telemetry with LKKs, which evolve to capture unquantifiable factors like individual gaming habits and software optimization levels.
• A macro-level analysis of 224 countries shows that GDP per capita and the Gini index explain 62.5% of the variability in a player's 95% FPS floor, indicating that regional economic factors significantly impact gaming performance.