Reward Design for a Reinforcement Learning Agent in an Android Environment

Reward Design for a Reinforcement Learning Agent in an Android Environment

Fauzi Sholichin

Software Engineer

Seroton Gmbh

Reward Design for a Reinforcement Learning Agent in an Android Environment

This talk introduces AndroidEnv, a Python library that turns Android devices into RL environments. It explores how to design reward systems in model-free RL, using algorithms like DQN and Q-learning to demonstrate how rewards shape an agent’s learning and goal achievement.