Adaptive discount factor for accelerating policy learning considering long-term returns in reinforcement learning with non-stationary environments

Puntos clave

• Adaptive discount factor enhances policy learning in reinforcement learning, especially in changing environments.
• A focus on long-term returns shows improved decision-making strategies across various scenarios.
• Observational analysis across environments reveals benefits for dynamic reinforcement learning applications.
• Highlights the need for continuous adaptation to optimize performance over time in changing circumstances.