Jun Wang

Yudi Zhang, Yali Du, Biwei Huang et al.

A major challenge in reinforcement learning is to determine which state-action pairs are responsible for future rewards that are delayed. Reward redistribution serves as a solution to re-assign credits for each time step from observed sequences. While the majority of current approaches construct the reward redistribution in an uninterpretable manner, we propose to explicitly model the contributions of state and action from a causal perspective, resulting in an interpretable reward redistribution and preserving policy invariance. In this paper, we start by studying the role of causal generative models in reward redistribution by characterizing the generation of Markovian rewards and trajectory-wise long-term return and further propose a framework, called Generative Return Decomposition (GRD), for policy optimization in delayed reward scenarios. Specifically, GRD first identifies the unobservable Markovian rewards and causal relations in the generative process. Then, GRD makes use of the identified causal generative model to form a compact representation to train policy over the most favorable subspace of the state space of the agent. Theoretically, we show that the unobservable Markovian reward function is identifiable, as well as the underlying causal structure and causal models. Experimental results show that our method outperforms state-of-the-art methods and the provided visualization further demonstrates the interpretability of our method. The project page is located at https://reedzyd.github.io/GenerativeReturnDecomposition/.

Minne Li, Zheng Tian, Pranav Nashikkar et al.

Existing model-based reinforcement learning methods often study perception modeling and decision making separately. We introduce joint Perception and Control as Inference (PCI), a general framework to combine perception and control for partially observable environments through Bayesian inference. Based on the fact that object-level inductive biases are critical in human perceptual learning and reasoning, we propose Object-based Perception Control (OPC), an instantiation of PCI which manages to facilitate control using automatic discovered object-based representations. We develop an unsupervised end-to-end solution and analyze the convergence of the perception model update. Experiments in a high-dimensional pixel environment demonstrate the learning effectiveness of our object-based perception control approach. Specifically, we show that OPC achieves good perceptual grouping quality and outperforms several strong baselines in accumulated rewards.

Rafał Muszyński, Jun Wang

Modeling personality is a challenging problem with applications spanning computer games, virtual assistants, online shopping and education. Many techniques have been tried, ranging from neural networks to computational cognitive architectures. However, most approaches rely on examples with hand-crafted features and scenarios. Here, we approach learning a personality by training agents using a Deep Q-Network (DQN) model on rewards based on psychoanalysis, against hand-coded AI in the game of Pong. As a result, we obtain 4 agents, each with its own personality. Then, we define happiness of an agent, which can be seen as a measure of alignment with agent's objective function, and study it when agents play both against hand-coded AI, and against each other. We find that the agents that achieve higher happiness during testing against hand-coded AI, have lower happiness when competing against each other. This suggests that higher happiness in testing is a sign of overfitting in learning to interact with hand-coded AI, and leads to worse performance against agents with different personalities.