We study the problem of training a Reinforcement Learning (RL) agent that is collaborative with humans without using human data. Although such agents can be obtained through self-play training, they can suffer significantly from the distributional shift when paired with unencountered partners, such as humans. In this paper, we propose Maximum Entropy Population-based training (MEP) to mitigate such distributional shift. In MEP, agents in the population are trained with our derived Population Entropy bonus to promote the pairwise diversity between agents and the individual diversity of agents themselves. After obtaining this diversified population, a common best agent is trained by paring with agents in this population via prioritized sampling, where the prioritization is dynamically adjusted based on the training progress. We demonstrate the effectiveness of our method MEP, with comparison to Self-Play PPO (SP), Population-Based Training (PBT), Trajectory Diversity (TrajeDi), and Fictitious Co-Play (FCP) in both matrix game and Overcooked game environments, with partners being human proxy models and real humans. A supplementary video showing experimental results is available at https://youtu.be/Xh-FKD0AAKE.

A large number of saliency methods have been proposed to explain individual decisions of deep convolutional neural networks (DCNNs). They work by identifying the relevance of each input feature to the predicted output class. However, the feature representations of hidden layers are difficult to interpret semantically. In this work, human-interpretable semantic concepts are associated with vectors in feature space. The association process is mathematically formulated as an optimization problem. The semantic vectors obtained from the optimal solution are applied to interpret deep neural networks globally and locally. The global interpretations are useful to understand the knowledge learned by DCNNs. The interpretation of local behaviors can help to gain a better understanding of the individual decisions made by DCNNs. The empirical experiments demonstrate how to use identified semantics to interpret the existing DCNNs.

In Multi-Goal Reinforcement Learning, an agent learns to achieve multiple goals with a goal-conditioned policy. During learning, the agent first collects the trajectories into a replay buffer, and later these trajectories are selected randomly for replay. However, the achieved goals in the replay buffer are often biased towards the behavior policies. From a Bayesian perspective, when there is no prior knowledge about the target goal distribution, the agent should learn uniformly from diverse achieved goals. Therefore, we first propose a novel multi-goal RL objective based on weighted entropy. This objective encourages the agent to maximize the expected return, as well as to achieve more diverse goals. Secondly, we developed a maximum entropy-based prioritization framework to optimize the proposed objective. For evaluation of this framework, we combine it with Deep Deterministic Policy Gradient, both with or without Hindsight Experience Replay. On a set of multi-goal robotic tasks of OpenAI Gym, we compare our method with other baselines and show promising improvements in both performance and sample-efficiency.

This paper is concerned with the training of recurrent neural networks as goal-oriented dialog agents using reinforcement learning. Training such agents with policy gradients typically requires a large amount of samples. However, the collection of the required data in form of conversations between chatbots and human agents is time-consuming and expensive. To mitigate this problem, we describe an efficient policy gradient method using positive memory retention, which significantly increases the sample-efficiency. We show that our method is 10 times more sample-efficient than policy gradients in extensive experiments on a new synthetic number guessing game. Moreover, in a real-word visual object discovery game, the proposed method is twice as sample-efficient as policy gradients and shows state-of-the-art performance.

Learning goal-oriented dialogues by means of deep reinforcement learning has recently become a popular research topic. However, commonly used policy-based dialogue agents often end up focusing on simple utterances and suboptimal policies. To mitigate this problem, we propose a class of novel temperature-based extensions for policy gradient methods, which are referred to as Tempered Policy Gradients (TPGs). On a recent AI-testbed, i.e., the GuessWhat?! game, we achieve significant improvements with two innovations. The first one is an extension of the state-of-the-art solutions with Seq2Seq and Memory Network structures that leads to an improvement of 7%. The second one is the application of our newly developed TPG methods, which improves the performance additionally by around 5% and, even more importantly, helps produce more convincing utterances.