Towards Better Interpretability in Deep Q-Networks
This addresses the need for better interpretability in deep reinforcement learning for researchers and practitioners, though it is incremental as it builds on existing Q-learning methods.
The paper tackled the problem of interpretability in deep Q-networks by proposing an interpretable neural network architecture that provides global explanations using key-value memories, attention, and reconstructible embeddings, achieving training rewards comparable to state-of-the-art models but revealing shallow features and overfitting issues in out-of-sample testing.
Deep reinforcement learning techniques have demonstrated superior performance in a wide variety of environments. As improvements in training algorithms continue at a brisk pace, theoretical or empirical studies on understanding what these networks seem to learn, are far behind. In this paper we propose an interpretable neural network architecture for Q-learning which provides a global explanation of the model's behavior using key-value memories, attention and reconstructible embeddings. With a directed exploration strategy, our model can reach training rewards comparable to the state-of-the-art deep Q-learning models. However, results suggest that the features extracted by the neural network are extremely shallow and subsequent testing using out-of-sample examples shows that the agent can easily overfit to trajectories seen during training.