Learning, Fast and Slow: A Goal-Directed Memory-Based Approach for Dynamic Environments
This work addresses slow training in reinforcement learning for dynamic environments, offering a novel hybrid method with strong empirical gains.
The paper tackles slow convergence in model-based reinforcement learning by introducing a dual-mechanism approach combining goal-directed exploration (fast) with memory-based planning (slow), achieving a 92% solve rate in a dynamic grid world, significantly outperforming PPO (54%), TRPO (50%), and A2C (24%).
Model-based next state prediction and state value prediction are slow to converge. To address these challenges, we do the following: i) Instead of a neural network, we do model-based planning using a parallel memory retrieval system (which we term the slow mechanism); ii) Instead of learning state values, we guide the agent's actions using goal-directed exploration, by using a neural network to choose the next action given the current state and the goal state (which we term the fast mechanism). The goal-directed exploration is trained online using hippocampal replay of visited states and future imagined states every single time step, leading to fast and efficient training. Empirical studies show that our proposed method has a 92% solve rate across 100 episodes in a dynamically changing grid world, significantly outperforming state-of-the-art actor critic mechanisms such as PPO (54%), TRPO (50%) and A2C (24%). Ablation studies demonstrate that both mechanisms are crucial. We posit that the future of Reinforcement Learning (RL) will be to model goals and sub-goals for various tasks, and plan it out in a goal-directed memory-based approach.