Pong

The 'Pong' dataset/benchmark contains a series of game scenarios used to evaluate the performance of reinforcement learning agents in learning effective policies while addressing issues related to spurious correlations and generalization in modified environments.

Papers using Pong (3)

Better Decisions through the Right Causal World Model2025Biological Neurons Compete with Deep Reinforcement Learning in Sample Efficiency in a Simulated Gameworld2024 · 3 citesRevealing the Learning Process in Reinforcement Learning Agents Through Attention-Oriented Metrics2024