Potential-based reward shaping provides an approach for designing good reward functions, with the purpose of speeding up learning. However, automatically finding potential functions for complex environments is a difficult problem (in fact, of the same difficulty as learning a value function from scratch). We propose a new framework for learning potential functions by leveraging ideas from graph representation learning. Our approach relies on Graph Convolutional Networks which we use as a key ingredient in combination with the probabilistic inference view of reinforcement learning. More precisely, we leverage Graph Convolutional Networks to perform message passing from rewarding states. The propagated messages can then be used as potential functions for reward shaping to accelerate learning. We verify empirically that our approach can achieve considerable improvements in both small and high-dimensional control problems.

The method adapts a sampled trajectory-based approximation of the transition graphs. But given the trajectory samples, sequential models (RNN etc.) are sufficient to estimate the potential functions. It would be good if the authors can clarify the advantage and the necessity of GCN on the sampled trajectory inputs, compared to sequential models. The baselines, ICM and RND, are motivated to address hard exploration RL tasks, while the potential based reward shaping is motivated for faster convergence. They are related but address different issues. A more informative empirical comparison would be against the LIRPG (Learning Intrinsic Rewards for Policy Gradient from [a]), because both this paper and LIRPG aim at learning reward shaping for speed up policy learning.

Potential-based reward shaping provides an approach for designing good reward functions, with the purpose of speeding up learning. However, automatically finding potential functions for complex environments is a difficult problem (in fact, of the same difficulty as learning a value function from scratch). We propose a new framework for learning potential functions by leveraging ideas from graph representation learning. Our approach relies on Graph Convolutional Networks which we use as a key ingredient in combination with the probabilistic inference view of reinforcement learning. More precisely, we leverage Graph Convolutional Networks to perform message passing from rewarding states.

In many real-world applications of control system and robotics, linear temporal logic (LTL) is a widely-used task specification language which has a compositional grammar that naturally induces temporally extended behaviours across tasks, including conditionals and alternative realizations. An important problem in RL with LTL tasks is to learn task-conditioned policies which can zero-shot generalize to new LTL instructions not observed in the training. However, because symbolic observation is often lossy and LTL tasks can have long time horizon, previous works can suffer from issues such as training sampling inefficiency and infeasibility or sub-optimality of the found solutions. In order to tackle these issues, this paper proposes a novel multi-task RL algorithm with improved learning efficiency and optimality. To achieve the global optimality of task completion, we propose to learn options dependent on the future subgoals via a novel off-policy approach. In order to propagate the rewards of satisfying future subgoals back more efficiently, we propose to train a multi-step value function conditioned on the subgoal sequence which is updated with Monte Carlo estimates of multi-step discounted returns. In experiments on three different domains, we evaluate the LTL generalization capability of the agent trained by the proposed method, showing its advantage over previous representative methods.

We propose a novel training algorithm for reinforcement learning which combines the strength of deep Q-learning with a constrained optimization approach to tighten optimality and encourage faster reward propagation. Our novel technique makes deep reinforcement learning more practical by drastically reducing the training time. We evaluate the performance of our approach on the 49 games of the challenging Arcade Learning Environment, and report significant improvements in both training time and accuracy.