Knowledge Transfer In Deep Reinforcement Learning Via An Rl-specific Gan-based Correspondence Function

Abstract

Deep reinforcement learning has demonstrated superhuman performance in complex decision-making tasks, but it struggles with generalization and knowledge reuse - key aspects of true intelligence. This article introduces a novel approach that modifies Cycle Generative Adversarial Networks specifically for reinforcement learning, enabling effective one-to-one knowledge transfer between two tasks. Our method enhances the loss function with two new components: model loss, which captures dynamic relationships between source and target tasks, and Q-loss, which identifies states significantly influencing the target decision policy. Tested on the 2-D Atari game Pong, our method achieved 100% knowledge transfer in identical tasks and either 100% knowledge transfer or a 30% reduction in training time for a rotated task, depending on the network architecture. In contrast, using standard Generative Adversarial Networks or Cycle Generative Adversarial Networks led to worse performance than training

Stats

• citations1
• S2 citations—
• github stars0
• HF likes0
• heat score2.26
• arxiv keyruman2022knowledge

Related papers

• Visual Transfer Between Atari Games Using Competitive Reinforcement Learning (2018)7.50
• Similarity-based Knowledge Transfer For Cross-domain Reinforcement Learning (2023)0.00
• Foundations For Transfer In Reinforcement Learning: A Taxonomy Of Knowledge Modalities (2023)0.00
• Probing Transfer In Deep Reinforcement Learning Without Task Engineering (2022)0.00
• Knowledge Transfer In Multi-task Deep Reinforcement Learning For Continuous Control (2020)0.00
• On The Transferability Of Deep-q Networks (2021)0.00
• Learning Of Generalizable And Interpretable Knowledge In Grid-based Reinforcement Learning Environments (2023)3.58
• SF-DQN: Provable Knowledge Transfer Using Successor Feature For Deep Reinforcement Learning (2024)0.00