We propose a new way of deriving policy gradient updates for reinforcement learning. Our technique, based on Fourier analysis, recasts integrals that arise with expected policy gradients as convolutions and turns them into multiplications. The obtained analytical solutions allow us to capture the low variance benefits of EPG in a broad range of settings. For the critic, we treat trigonometric and radial basis functions, two function families with the universal approximation property. The choice of policy can be almost arbitrary, including mixtures or hybrid continuous-discrete probability distributions. Moreover, we derive a general family of sample-based estimators for stochastic policy gradients, which unifies existing results on sample-based approximation. We believe that this technique has the potential to shape the next generation of policy gradient approaches, powered by analytical results.

Animals are able to reach a desired state in an environment by controlling various behavioral patterns. Identification of the behavioral strategy used for this control is important for understanding animals' decision-making and is fundamental to dissect information processing done by the nervous system. However, methods for quantifying such behavioral strategies have not been fully established. In this study, we developed an inverse reinforcement-learning (IRL) framework to identify an animal's behavioral strategy from behavioral time-series data. We applied this framework to C. elegans thermotactic behavior; after cultivation at a constant temperature with or without food, fed worms prefer, while starved worms avoid the cultivation temperature on a thermal gradient.

In this paper, we propose to combine imitation and reinforcement learning via the idea of reward shaping using an oracle. We study the effectiveness of the near-optimal cost-to-go oracle on the planning horizon and demonstrate that the cost-to-go oracle shortens the learner's planning horizon as function of its accuracy: a globally optimal oracle can shorten the planning horizon to one, leading to a one-step greedy Markov Decision Process which is much easier to optimize, while an oracle that is far away from the optimality requires planning over a longer horizon to achieve near-optimal performance. Hence our new insight bridges the gap and interpolates between imitation learning and reinforcement learning. Motivated by the above mentioned insights, we propose Truncated HORizon Policy Search (THOR), a method that focuses on searching for policies that maximize the total reshaped reward over a finite planning horizon when the oracle is sub-optimal. We experimentally demonstrate that a gradient-based implementation of THOR can achieve superior performance compared to RL baselines and IL baselines even when the oracle is sub-optimal.

Model-based deep reinforcement learning (DRL) algorithm uses the sampled data from a real environment to learn the underlying system dynamics to construct an approximate cyber environment. By using the synthesized data generated from the cyber environment to train the target controller, the training cost can be reduced significantly. In current research, issues such as the applicability of approximate model and the strategy to sample and train from the real and cyber environment have not been fully investigated. To address these issues, we propose to utilize an intelligent trainer to properly use the approximate model and control the sampling and training procedure in the model-based DRL. To do so, we package the training process of a model-based DRL as a standard RL environment, and design an RL trainer to control the training process. The trainer has three control actions: the first action controls where to sample in the real and cyber environment; the second action determines how many data should be sampled from the cyber environment and the third action controls how many times the cyber data should be used to train the target controller. These actions would be controlled manually if without the trainer. The proposed framework is evaluated on five different tasks of OpenAI gym and the test results show that the proposed trainer achieves significant better performance than a fixed parameter model-based RL baseline algorithm. In addition, we compare the performance of the intelligent trainer to a random trainer and prove that the intelligent trainer can indeed learn on the fly. The proposed training framework can be extended to more control actions with more sophisticated trainer design to further reduce the tweak cost of model-based RL algorithms.

We propose a new sample-efficient methodology, called Supervised Policy Update (SPU), for deep reinforcement learning. Starting with data generated by the current policy, SPU optimizes over the proximal policy space to find a non-parameterized policy. It then solves a supervised regression problem to convert the non-parameterized policy to a parameterized policy, from which it draws new samples. There is significant flexibility in setting the labels in the supervised regression problem, with different settings corresponding to different underlying optimization problems. We develop a methodology for finding an optimal policy in the non-parameterized policy space, and show how Trust Region Policy Optimization (TRPO) and Proximal Policy Optimization (PPO) can be addressed by this methodology. In terms of sample efficiency, our experiments show SPU can outperform PPO for simulated robotic locomotion tasks.

Reinforcement learning (RL) is a systematic approach to learning in sequential decision problems, where a learners' future task performance depends on its past actions. In such settings, learners have to explore, meaning they have to take actions with uncertain outcomes, to facilitate learning about the consequences of such actions. The question of how to best explore is a key open question in RL. Here, we specifically address this question from an empirical perspective, and investigate how to explore in a way that leads to sample efficient learning in deep RL, i.e., reinforcement learning with value function approximators that are parameterized as powerful neural networks. We present a surprising finding: in this setting, good approximate value functions can be learned without any explicit exploration. In fact, we find that in several cases learning without explicit exploration is equally or more sample efficient than the most-commonly used ɛ-greedy exploration scheme on several standard benchmark tasks. We present additional results that suggest a likely role of model structure (network depth and nonlinearity) in inducing such implicit exploration. We believe that our insights have strong practical implications and open up a novel line of research towards understanding exploration in deep RL.

Despite significant advances in the field of deep Reinforcement Learning (RL), today's algorithms still fail to learn human-level policies consistently over a set of diverse tasks such as Atari 2600 games. We identify three key challenges that any algorithm needs to master in order to perform well on all games: processing diverse reward distributions, reasoning over long time horizons, and exploring efficiently. In this paper, we propose an algorithm that addresses each of these challenges and is able to learn human-level policies on nearly all Atari games. A new transformed Bellman operator allows our algorithm to process rewards of varying densities and scales; an auxiliary temporal consistency loss allows us to train stably using a discount factor of $\gamma = 0.999$ (instead of $\gamma = 0.99$) extending the effective planning horizon by an order of magnitude; and we ease the exploration problem by using human demonstrations that guide the agent towards rewarding states. When tested on a set of 42 Atari games, our algorithm exceeds the performance of an average human on 40 games using a common set of hyper parameters. Furthermore, it is the first deep RL algorithm to solve the first level of Montezuma's Revenge.

Deep reinforcement learning methods traditionally struggle with tasks where environment rewards are particularly sparse. One successful method of guiding exploration in these domains is to imitate trajectories provided by a human demonstrator. However, these demonstrations are typically collected under artificial conditions, i.e. with access to the agent's exact environment setup and the demonstrator's action and reward trajectories. Here we propose a two-stage method that overcomes these limitations by relying on noisy, unaligned footage without access to such data. First, we learn to map unaligned videos from multiple sources to a common representation using self-supervised objectives constructed over both time and modality (i.e. vision and sound). Second, we embed a single YouTube video in this representation to construct a reward function that encourages an agent to imitate human gameplay. This method of one-shot imitation allows our agent to convincingly exceed human-level performance on the infamously hard exploration games Montezuma's Revenge, Pitfall! and Private Eye for the first time, even if the agent is not presented with any environment rewards.

We show that when a third party, the adversary, steps into the two-party setting (agent and operator) of safely interruptible reinforcement learning, a trade-off has to be made between the probability of following the optimal policy in the limit, and the probability of escaping a dangerous situation created by the adversary. So far, the work on safely interruptible agents has assumed a perfect perception of the agent about its environment (no adversary), and therefore implicitly set the second probability to zero, by explicitly seeking a value of one for the first probability. We show that (1) agents can be made both interruptible and adversary-resilient, and (2) the interruptibility can be made safe in the sense that the agent itself will not seek to avoid it. We also solve the problem that arises when the agent does not go completely greedy, i.e. issues with safe exploration in the limit. Resilience to perturbed perception, safe exploration in the limit, and safe interruptibility are the three pillars of what we call \emph{virtuously safe reinforcement learning}.

We consider the transfer of experience samples (i.e., tuples < s, a, s', r >) in reinforcement learning (RL), collected from a set of source tasks to improve the learning process in a given target task. Most of the related approaches focus on selecting the most relevant source samples for solving the target task, but then all the transferred samples are used without considering anymore the discrepancies between the task models. In this paper, we propose a model-based technique that automatically estimates the relevance (importance weight) of each source sample for solving the target task. In the proposed approach, all the samples are transferred and used by a batch RL algorithm to solve the target task, but their contribution to the learning process is proportional to their importance weight. By extending the results for importance weighting provided in supervised learning literature, we develop a finite-sample analysis of the proposed batch RL algorithm. Furthermore, we empirically compare the proposed algorithm to state-of-the-art approaches, showing that it achieves better learning performance and is very robust to negative transfer, even when some source tasks are significantly different from the target task.