🤖 AI Summary
In sparse-reward environments, standard DQN suffers from unstable convergence and low learning efficiency due to its fixed learning rate and uniform update mechanism. This work proposes DISRC, a novel approach that integrates a biologically inspired surprise mechanism into DQN. DISRC employs a LayerNorm encoder to construct state representations and computes a discrepancy-based intrinsic surprise signal derived from a moving latent variable setpoint. This surprise signal is combined with the temporal difference (TD) error to dynamically modulate the intensity of Q-value updates, enabling expectation-violation-driven adaptive learning. Evaluated on MiniGrid-DoorKey, DISRC achieves a 33% faster training speed, lower reward variance, and higher AUC. On LavaCrossing, it attains the best-reported AUC (957.04) and final reward, demonstrating significantly improved exploration efficiency and training stability.
📝 Abstract
Deep reinforcement learning (DRL) has driven major advances in autonomous control. Still, standard Deep Q-Network (DQN) agents tend to rely on fixed learning rates and uniform update scaling, even as updates are modulated by temporal-difference (TD) error. This rigidity destabilizes convergence, especially in sparse-reward settings where feedback is infrequent. We introduce Deep Intrinsic Surprise-Regularized Control (DISRC), a biologically inspired augmentation to DQN that dynamically scales Q-updates based on latent-space surprise. DISRC encodes states via a LayerNorm-based encoder and computes a deviation-based surprise score relative to a moving latent setpoint. Each update is then scaled in proportion to both TD error and surprise intensity, promoting plasticity during early exploration and stability as familiarity increases. We evaluate DISRC on two sparse-reward MiniGrid environments, which included MiniGrid-DoorKey-8x8 and MiniGrid-LavaCrossingS9N1, under identical settings as a vanilla DQN baseline. In DoorKey, DISRC reached the first successful episode (reward>0.8) 33% faster than the vanilla DQN baseline (79 vs. 118 episodes), with lower reward standard deviation (0.25 vs. 0.34) and higher reward area under the curve (AUC: 596.42 vs. 534.90). These metrics reflect faster, more consistent learning - critical for sparse, delayed reward settings. In LavaCrossing, DISRC achieved a higher final reward (0.95 vs. 0.93) and the highest AUC of all agents (957.04), though it converged more gradually. These preliminary results establish DISRC as a novel mechanism for regulating learning intensity in off-policy agents, improving both efficiency and stability in sparse-reward domains. By treating surprise as an intrinsic learning signal, DISRC enables agents to modulate updates based on expectation violations, enhancing decision quality when conventional value-based methods fall short.