Monte Carlo sampling in high-dimensional, low-sample settings is important in many machine learning tasks. We improve current methods for sampling in Euclidean spaces by avoiding independence, and instead consider ways to couple samples. We show fundamental connections to optimal transport theory, leading to novel sampling algorithms, and providing new theoretical grounding for existing strategies. We compare our new strategies against prior methods for improving sample efficiency, including QMC, by studying discrepancy. We explore our findings empirically, and observe benefits of our sampling schemes for reinforcement learning and generative modelling.

For an autonomous agent to fulfill a wide range of user-specified goals at test time, it must be able to learn broadly applicable and general-purpose skill repertoires. Furthermore, to provide the requisite level of generality, these skills must handle raw sensory input such as images. In this paper, we propose an algorithm that acquires such general-purpose skills by combining unsupervised representation learning and reinforcement learning of goal-conditioned policies. Since the particular goals that might be required at test-time are not known in advance, the agent performs a self-supervised "practice" phase where it imagines goals and attempts to achieve them. We learn a visual representation with three distinct purposes: sampling goals for self-supervised practice, providing a structured transformation of raw sensory inputs, and computing a reward signal for goal reaching.

Learning how to act when there are many available actions in each state is a challenging task for Reinforcement Learning (RL) agents, especially when many of the actions are redundant or irrelevant. In such cases, it is easier to learn which actions not to take. In this work, we propose the Action-Elimination Deep Q-Network (AE-DQN) architecture that combines a Deep RL algorithm with an Action Elimination Network (AEN) that eliminates sub-optimal actions. The AEN is trained to predict invalid actions, supervised by an external elimination signal provided by the environment. Simulations demonstrate a considerable speedup and added robustness over vanilla DQN in text-based games with over a thousand discrete actions.

It is commonly believed that an agent making decisions on behalf of two or more principals who have different utility functions should adopt a Pareto optimal policy, i.e. a policy that cannot be improved upon for one principal without making sacrifices for another. Harsanyi's theorem shows that when the principals have a common prior on the outcome distributions of all policies, a Pareto optimal policy for the agent is one that maximizes a fixed, weighted linear combination of the principals' utilities. In this paper, we derive a more precise generalization for the sequential decision setting in the case of principals with different priors on the dynamics of the environment. We refer to this generalization as the Negotiable Reinforcement Learning (NRL) framework. In this more general case, the relative weight given to each principal's utility should evolve over time according to how well the agent's observations conform with that principal's prior.

The goal of reinforcement learning algorithms is to estimate and/or optimise the value function. However, unlike supervised learning, no teacher or oracle is available to provide the true value function. Instead, the majority of reinforcement learning algorithms estimate and/or optimise a proxy for the value function. This proxy is typically based on a sampled and bootstrapped approximation to the true value function, known as a return. The particular choice of return is one of the chief components determining the nature of the algorithm: the rate at which future rewards are discounted; when and how values should be bootstrapped; or even the nature of the rewards themselves.

Learning to cooperate with friends and compete with foes is a key component of multi-agent reinforcement learning. Typically to do so, one requires access to either a model of or interaction with the other agent(s). Here we show how to learn effective strategies for cooperation and competition in an asymmetric information game with no such model or interaction. Our approach is to encourage an agent to reveal or hide their intentions using an information-theoretic regularizer. We consider both the mutual information between goal and action given state, as well as the mutual information between goal and state.

SUMMARY: The paper considers entropy regularized discounted Markov Decision Process (MDP), and shows the relation between the optimal value, action-value, and policy. Moreover, it shows that the optimal value function and policy satisfy a temporal consistency in the form of Bellman-like equation (Theorem 1), which can also be extended to its n-step version (Corollary 2). The paper introduces Path Consistent Learning by enforcing the temporal consistency, which is essentially a Bellman residual minimization procedure (Section 5). SUMMARY OF EVALUATION: Quality: Parts of the paper is sound (Section 3 and 4); parts of it is not (Section 5) Clarity: The paper is well-written. Originality: Some results seem to be novel, but similar ideas and analysis have been proposed/done before.

The paper presents a new method for structured output prediction using reinforcement learning. Previous methods used reward augmented maximum likelihoods or policy gradients. The new method uses a soft-max objective. The authors present a new inference method that can be used to efficiently evaluate the integral in the objective. In addition, the authors propose to use additional reward functions which encode prior knowledge (e.g. to avoid word repetitions).

This paper proposes an approach for behavioral cloning that constructs a function space for a particular parametric policy model based on the null space of the policy gradient. I think a running example (e.g., for discrete MDP) would help explain the approach. I found myself flipping back and forth from the Algorithm (page 6) to the description of each step. I have some lingering confusion about using Eq. I assume a similar estimator is employed for d(s,a).

Summary: This paper proposes a method that can help an RL agent to rapidly adapt to experience in the replay buffer. The method is the combination of slow and general component (i.e. An interesting part of this proposed approach is that it slightly changes the replay buffer by adding trajectory information but get a good boost in the perfoamnce. In addition, the extensive number of experiments have been conducted in order to verify the claim of this paper. Comments and Questions - This paper, in general, is well-written (especially the related works, they actually talk about the relation and difference with previous works) except for the followings: -- Section 3.1 and 3.2 do not have smooth flow.