Learning, Fast and Slow: A Goal-Directed Memory-Based Approach for Dynamic Environments

TL;DR

This paper introduces a goal-directed, memory-based reinforcement learning approach that combines a slow, model-based planning mechanism with a fast, goal-guided exploration strategy, achieving high success rates in dynamic environments.

Contribution

The paper proposes a novel dual-mechanism framework that integrates memory-based planning with goal-directed exploration, improving learning speed and performance in dynamic settings.

Findings

92% solve rate in dynamic grid world

Outperforms PPO, TRPO, A2C

Both mechanisms are essential for success

Abstract

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.