Model-based reinforcement learning (MBRL) has shown its advantages in sample-efficiency over model-free reinforcement learning (MFRL). Despite the impressive results it achieves, it still faces a trade-off between the ease of data generation and model bias. In this paper, we propose a simple and elegant model-embedding model-based reinforcement learning (MEMB) algorithm in the framework of the probabilistic reinforcement learning. To balance the sample-efficiency and model bias, we exploit both real and imaginary data in the training. In particular, we embed the model in the policy update and learn $Q$ and $V$ functions from the real data set. We provide the theoretical analysis of MEMB with the Lipschitz continuity assumption on the model and policy. At last, we evaluate MEMB on several benchmarks and demonstrate our algorithm can achieve state-of-the-art performance.

Learning value functions off-policy is at the core of modern Reinforcement Learning (RL). Traditional off-policy actor-critic algorithms, however, only approximate the true policy gradient, since the gradient $\nabla_{\theta} Q^{\pi_{\theta}}(s,a)$ of the action-value function with respect to the policy parameters is often ignored. We introduce a class of value functions called Parameter-based Value Functions (PVFs) whose inputs include the policy parameters. PVFs can evaluate the performance of any policy given a state, a state-action pair, or a distribution over the RL agent's initial states. We show how PVFs yield exact policy gradient theorems. We derive off-policy actor-critic algorithms based on PVFs trained using Monte Carlo or Temporal Difference methods. Preliminary experimental results indicate that PVFs can effectively evaluate deterministic linear and nonlinear policies, outperforming state-of-the-art algorithms in the continuous control environment Swimmer-v3. Finally, we show how recurrent neural networks can be trained through PVFs to solve supervised and RL problems involving partial observability and long time lags between relevant events. This provides an alternative to backpropagation through time.

Continually solving new, unsolved tasks is the key to learning diverse behaviors. Through reinforcement learning (RL), we have made massive strides towards solving tasks that have a single goal. However, in the multi-task domain, where an agent needs to reach multiple goals, the choice of training goals can largely affect sample efficiency. When biological agents learn, there is often an organized and meaningful order to which learning happens. Inspired by this, we propose setting up an automatic curriculum for goals that the agent needs to solve. Our key insight is that if we can sample goals at the frontier of the set of goals that an agent is able to reach, it will provide a significantly stronger learning signal compared to randomly sampled goals. To operationalize this idea, we introduce a goal proposal module that prioritizes goals that maximize the epistemic uncertainty of the Q-function of the policy. This simple technique samples goals that are neither too hard nor too easy for the agent to solve, hence enabling continual improvement. We evaluate our method across 13 multi-goal robotic tasks and 5 navigation tasks, and demonstrate performance gains over current state-of-the-art methods.

Network densification and millimeter-wave technologies are key enablers to fulfill the capacity and data rate requirements of the fifth generation (5G) of mobile networks. In this context, designing low-complexity policies with local observations, yet able to adapt the user association with respect to the global network state and to the network dynamics is a challenge. In fact, the frameworks proposed in literature require continuous access to global network information and to recompute the association when the radio environment changes. With the complexity associated to such an approach, these solutions are not well suited to dense 5G networks. In this paper, we address this issue by designing a scalable and flexible algorithm for user association based on multi-agent reinforcement learning. In this approach, users act as independent agents that, based on their local observations only, learn to autonomously coordinate their actions in order to optimize the network sum-rate. Since there is no direct information exchange among the agents, we also limit the signaling overhead. Simulation results show that the proposed algorithm is able to adapt to (fast) changes of radio environment, thus providing large sum-rate gain in comparison to state-of-the-art solutions.

Efficient exploration is one of the main challenges in reinforcement learning (RL). Most existing sample-efficient algorithms assume the existence of a single reward function during exploration. In many practical scenarios, however, there is not a single underlying reward function to guide the exploration, for instance, when an agent needs to learn many skills simultaneously, or multiple conflicting objectives need to be balanced. To address these challenges, we propose the \textit{task-agnostic RL} framework: In the exploration phase, the agent first collects trajectories by exploring the MDP without the guidance of a reward function. After exploration, it aims at finding near-optimal policies for $N$ tasks, given the collected trajectories augmented with \textit{sampled rewards} for each task. We present an efficient task-agnostic RL algorithm, \textsc{UCBZero}, that finds $\epsilon$-optimal policies for $N$ arbitrary tasks after at most $\tilde O(\log(N)H^5SA/\epsilon^2)$ exploration episodes. We also provide an $\Omega(\log (N)H^2SA/\epsilon^2)$ lower bound, showing that the $\log$ dependency on $N$ is unavoidable. Furthermore, we provide an $N$-independent sample complexity bound of \textsc{UCBZero} in the statistically easier setting when the ground truth reward functions are known.

Learning effective contextual-bandit policies from past actions of a deployed system is highly desirable in many settings (e.g. voice assistants, recommendation, search), since it enables the reuse of large amounts of log data. State-of-the-art methods for such off-policy learning, however, are based on inverse propensity score (IPS) weighting. A key theoretical requirement of IPS weighting is that the policy that logged the data has "full support", which typically translates into requiring non-zero probability for any action in any context. Unfortunately, many real-world systems produce support deficient data, especially when the action space is large, and we show how existing methods can fail catastrophically. To overcome this gap between theory and applications, we identify three approaches that provide various guarantees for IPS-based learning despite the inherent limitations of support-deficient data: restricting the action space, reward extrapolation, and restricting the policy space. We systematically analyze the statistical and computational properties of these three approaches, and we empirically evaluate their effectiveness. In addition to providing the first systematic analysis of support-deficiency in contextual-bandit learning, we conclude with recommendations that provide practical guidance.

Q-learning [Watkins and Dayan, 1992] is one of the most popular classes of methods for solving reinforcement learning (RL) problems. Q-learning tries to estimate the optimal state-action value function (Q-function). With a Q-function, at every state, one can greedily choose the action with the largest Q value to interact with the RL environment while achieving near optimal expected cumulative rewards in the long run. Compared to another popular classes of methods, e.g., modelbased RL, Q-learning algorithms (or more generally, model-free algorithms) often enjoy better memory and time efficiency

Meta reinforcement learning (meta-RL) provides a principled approach for fast adaptation to novel tasks by extracting prior knowledge from previous tasks. Under such settings, it is crucial for the agent to perform efficient exploration during adaptation to collect useful experiences. However, existing methods suffer from poor adaptation performance caused by inefficient exploration mechanisms, especially in sparse-reward problems. In this paper, we present a novel off-policy context-based meta-RL approach that efficiently learns a separate exploration policy to support fast adaptation, as well as a context-aware exploitation policy to maximize extrinsic return. The explorer is motivated by an information-theoretical intrinsic reward that encourages the agent to collect experiences that provide rich information about the task. Experiment results on both MuJoCo and Meta-World benchmarks show that our method significantly outperforms baselines by performing efficient exploration strategies.

We propose a novel reinforcement learning algorithm,QD-RL, that incorporates the strengths of off-policy RL algorithms into Quality Diversity (QD) approaches. Quality-Diversity methods contribute structural biases by decoupling the search for diversity from the search for high return, resulting in efficient management of the exploration-exploitation trade-off. However, these approaches generally suffer from sample inefficiency as they call upon evolutionary techniques. QD-RL removes this limitation by relying on off-policy RL algorithms. More precisely, we train a population of off-policy deep RL agents to simultaneously maximize diversity inside the population and the return of the agents. QD-RL selects agents from the diversity-return Pareto Front, resulting in stable and efficient population updates. Our experiments on the Ant-Maze environment show that QD-RL can solve challenging exploration and control problems with deceptive rewards while being more than 15 times more sample efficient than its evolutionary counterparts.

Meta-reinforcement learning (meta-RL) aims to learn from multiple training tasks the ability to adapt efficiently to unseen test tasks. Despite the success, existing meta-RL algorithms are known to be sensitive to the task distribution shift. When the test task distribution is different from the training task distribution, the performance may degrade significantly. To address this issue, this paper proposes Model-based Adversarial Meta-Reinforcement Learning (AdMRL), where we aim to minimize the worst-case sub-optimality gap -- the difference between the optimal return and the return that the algorithm achieves after adaptation -- across all tasks in a family of tasks, with a model-based approach. We propose a minimax objective and optimize it by alternating between learning the dynamics model on a fixed task and finding the adversarial task for the current model -- the task for which the policy induced by the model is maximally suboptimal. Assuming the family of tasks is parameterized, we derive a formula for the gradient of the suboptimality with respect to the task parameters via the implicit function theorem, and show how the gradient estimator can be efficiently implemented by the conjugate gradient method and a novel use of the REINFORCE estimator. We evaluate our approach on several continuous control benchmarks and demonstrate its efficacy in the worst-case performance over all tasks, the generalization power to out-of-distribution tasks, and in training and test time sample efficiency, over existing state-of-the-art meta-RL algorithms.