Humans show an innate ability to learn the regularities of the world through interaction. By performing experiments in our environment, we are able to discern the causal factors of variation and infer how they affect the dynamics of our world. Analogously, here we attempt to equip reinforcement learning agents with the ability to perform experiments that facilitate a categorization of the rolled-out trajectories, and to subsequently infer the causal factors of the environment in a hierarchical manner. We introduce a novel intrinsic reward, called causal curiosity, and show that it allows our agents to learn optimal sequences of actions, and to discover causal factors in the dynamics. The learned behavior allows the agent to infer a binary quantized representation for the ground-truth causal factors in every environment. Additionally, we find that these experimental behaviors are semantically meaningful (e.g., to differentiate between heavy and light blocks, our agents learn to lift them), and are learnt in a self-supervised manner with approximately 2.5 times less data than conventional supervised planners. We show that these behaviors can be re-purposed and fine-tuned (e.g., from lifting to pushing or other downstream tasks). Finally, we show that the knowledge of causal factor representations aids zero-shot learning for more complex tasks.

As reinforcement learning agents become increasingly integrated into complex, real-world environments, designing for safety becomes a critical consideration. We specifically focus on researching scenarios where agents can cause undesired side effects while executing a policy on a primary task. Since one can define multiple tasks for a given environment dynamics, there are two important challenges. First, we need to abstract the concept of safety that applies broadly to that environment independent of the specific task being executed. Second, we need a mechanism for the abstracted notion of safety to modulate the actions of agents executing different policies to minimize their side-effects. In this work, we propose Safety Aware Reinforcement Learning (SARL) - a framework where a virtual safe agent modulates the actions of a main reward-based agent to minimize side effects. The safe agent learns a task-independent notion of safety for a given environment. The main agent is then trained with a regularization loss given by the distance between the native action probabilities of the two agents. Since the safe agent effectively abstracts a task-independent notion of safety via its action probabilities, it can be ported to modulate multiple policies solving different tasks within the given environment without further training. We contrast this with solutions that rely on task-specific regularization metrics and test our framework on the SafeLife Suite, based on Conway's Game of Life, comprising a number of complex tasks in dynamic environments. We show that our solution is able to match the performance of solutions that rely on task-specific side-effect penalties on both the primary and safety objectives while additionally providing the benefit of generalizability and portability.

In this paper, we study the problem of autonomously discovering temporally abstracted actions, or options, for exploration in reinforcement learning. For learning diverse options suitable for exploration, we introduce the infomax termination objective defined as the mutual information between options and their corresponding state transitions. We derive a scalable optimization scheme for maximizing this objective via the termination condition of options, yielding the InfoMax Option Critic (IMOC) algorithm. Through illustrative experiments, we empirically show that IMOC learns diverse options and utilizes them for exploration. Moreover, we show that IMOC scales well to continuous control tasks.

Reinforcement learning in heterogeneous multi-agent scenarios is important for real-world applications but presents challenges beyond those seen in homogeneous settings and simple benchmarks. In this work, we present an actor-critic algorithm that allows a team of heterogeneous agents to learn decentralized control policies for covering an unknown environment. This task is of interest to national security and emergency response organizations that would like to enhance situational awareness in hazardous areas by deploying teams of unmanned aerial vehicles. To solve this multi-agent coverage path planning problem in unknown environments, we augment a multi-agent actor-critic architecture with a new state encoding structure and triplet learning loss to support heterogeneous agent learning. We developed a simulation environment that includes real-world environmental factors such as turbulence, delayed communication, and agent loss, to train teams of agents as well as probe their robustness and flexibility to such disturbances.

An effective approach to exploration in reinforcement learning is to rely on an agent's uncertainty over the optimal policy, which can yield near-optimal exploration strategies in tabular settings. However, in non-tabular settings that involve function approximators, obtaining accurate uncertainty estimates is almost as challenging a problem. In this paper, we highlight that value estimates are easily biased and temporally inconsistent. In light of this, we propose a novel method for estimating uncertainty over the value function that relies on inducing a distribution over temporal difference errors. This exploration signal controls for state-action transitions so as to isolate uncertainty in value that is due to uncertainty over the agent's parameters. Instead, we incorporate it as an intrinsic reward and treat exploration as a separate learning problem, induced by the agent's temporal difference uncertainties. We introduce a distinct exploration policy that learns to collect data with high estimated uncertainty, which gives rise to a "curriculum" that smoothly changes throughout learning and vanishes in the limit of perfect value estimates. We evaluate our method on hard-exploration tasks, including Deep Sea and Atari 2600 environments and find that our proposed form of exploration facilitates both diverse and deep exploration. Striking the right balance between exploration and exploitation is fundamental to the reinforcement learning problem. A common approach is to derive exploration from the policy being learned. Dithering strategies, such as ɛ-greedy exploration, render a reward-maximising policy stochastic around its reward maximising behaviour (Williams & Peng, 1991). Other methods encourage higher entropy in the policy (Ziebart et al., 2008), introduce an intrinsic reward (Singh et al., 2005), or drive exploration by sampling from the agent's belief over the MDP (Strens, 2000). While greedy or entropy-maximising policies cannot facilitate temporally extended exploration (Osband et al., 2013; 2016a), the efficacy of intrinsic rewards depends crucially on how they relate to the extrinsic reward that comes from the environment (Burda et al., 2018a).

The training of autonomous agents often requires expensive and unsafe trial-and-error interactions with the environment. Nowadays several data sets containing recorded experiences of intelligent agents performing various tasks, spanning from the control of unmanned vehicles to human-robot interaction and medical applications are accessible on the internet. With the intention of limiting the costs of the learning procedure it is convenient to exploit the information that is already available rather than collecting new data. Nevertheless, the incapability to augment the batch can lead the autonomous agents to develop far from optimal behaviours when the sampled experiences do not allow for a good estimate of the true distribution of the environment. Offline learning is the area of machine learning concerned with efficiently obtaining an optimal policy with a batch of previously collected experiences without further interaction with the environment. In this paper we adumbrate the ideas motivating the development of the state-of-the-art offline learning baselines. The listed methods consist in the introduction of epistemic uncertainty dependent constraints during the classical resolution of a Markov Decision Process, with and without function approximators, that aims to alleviate the bad effects of the distributional mismatch between the available samples and real world. We provide comments on the practical utility of the theoretical bounds that justify the application of these algorithms and suggest the utilization of Generative Adversarial Networks to estimate the distributional shift that affects all of the proposed model-free and model-based approaches.

Reinforcement learning (RL) methods usually treat reward functions as black boxes. As such, these methods must extensively interact with the environment in order to discover rewards and optimal policies. In most RL applications, however, users have to program the reward function and, hence, there is the opportunity to treat reward functions as white boxes instead -- to show the reward function's code to the RL agent so it can exploit its internal structures to learn optimal policies faster. In this paper, we show how to accomplish this idea in two steps. First, we propose reward machines (RMs), a type of finite state machine that supports the specification of reward functions while exposing reward function structure. We then describe different methodologies to exploit such structures, including automated reward shaping, task decomposition, and counterfactual reasoning for data augmentation. Experiments on tabular and continuous domains show the benefits of exploiting reward structure across different tasks and RL agents.

Multi-Agent Reinforcement Learning (MARL) has demonstrated significant success in training decentralised policies in a centralised manner by making use of value factorization methods. However, addressing surprise across spurious states and approximation bias remain open problems for multi-agent settings. We introduce the Energy-based MIXer (EMIX), an algorithm which minimizes surprise utilizing the energy across agents. Our contributions are threefold; (1) EMIX introduces a novel surprise minimization technique across multiple agents in the case of multi-agent partially-observable settings. (2) EMIX highlights the first practical use of energy functions in MARL (to our knowledge) with theoretical guarantees and experiment validations of the energy operator. Lastly, (3) EMIX presents a novel technique for addressing overestimation bias across agents in MARL. When evaluated on a range of challenging StarCraft II micromanagement scenarios, EMIX demonstrates consistent state-of-the-art performance for multi-agent surprise minimization. Moreover, our ablation study highlights the necessity of the energy-based scheme and the need for elimination of overestimation bias in MARL. Our implementation of EMIX and videos of agents are available at https://karush17.github.io/emix-web/.

Despite significant progress in challenging problems across various domains, applying state-of-the-art deep reinforcement learning (RL) algorithms remains challenging due to their sensitivity to the choice of hyperparameters. This sensitivity can partly be attributed to the non-stationarity of the RL problem, potentially requiring different hyperparameter settings at various stages of the learning process. Additionally, in the RL setting, hyperparameter optimization (HPO) requires a large number of environment interactions, hindering the transfer of the successes in RL to real-world applications. In this work, we tackle the issues of sample-efficient and dynamic HPO in RL. We propose a population-based automated RL (AutoRL) framework to meta-optimize arbitrary off-policy RL algorithms. By sharing the collected experience across the population, we substantially increase the sample efficiency of the meta-optimization. We demonstrate the capabilities of our sample-efficient AutoRL approach in a case study with the popular TD3 algorithm in the MuJoCo benchmark suite, where we reduce the number of environment interactions needed for meta-optimization by up to an order of magnitude compared to population-based training. Deep reinforcement learning (RL) algorithms are often sensitive to the choice of internal hyperparameters (Jaderberg et al., 2017; Mahmood et al., 2018), and the hyperparameters of the neural network architecture (Islam et al., 2017; Henderson et al., 2018), hindering them from being applied out-of-the-box to new environments. Tuning hyperparameters of RL algorithms can quickly become very expensive, both in terms of high computational costs and a large number of required environment interactions. Especially in real-world applications, sample efficiency is crucial (Lee et al., 2019). Hyperparameter optimization (HPO; Snoek et al., 2012; Feurer & Hutter, 2019) approaches often treat the algorithm under optimization as a black-box, which in the setting of RL requires a full training run every time a configuration is evaluated. This leads to a suboptimal sample efficiency in terms of environment interactions.

QTRAN is a multi-agent reinforcement learning (MARL) algorithm capable of learning the largest class of joint-action value functions up to date. However, despite its strong theoretical guarantee, it has shown poor empirical performance in complex environments, such as Starcraft Multi-Agent Challenge (SMAC). In this paper, we identify the performance bottleneck of QTRAN and propose a substantially improved version, coined QTRAN++. Our gains come from (i) stabilizing the training objective of QTRAN, (ii) removing the strict role separation between the action-value estimators of QTRAN, and (iii) introducing a multi-head mixing network for value transformation. Through extensive evaluation, we confirm that our diagnosis is correct, and QTRAN++ successfully bridges the gap between empirical performance and theoretical guarantee. In particular, QTRAN++ newly achieves state-of-the-art performance in the SMAC environment. The code will be released.