In this paper we present a world model, which learns causal features using the invariance principle. In particular, we use contrastive unsupervised learning to learn the invariant causal features, which enforces invariance across augmentations of irrelevant parts or styles of the observation. The world-model-based reinforcement learning methods independently optimize representation learning and the policy. Thus naive contrastive loss implementation collapses due to a lack of supervisory signals to the representation learning module. We propose an intervention invariant auxiliary task to mitigate this issue. Specifically, we utilize depth prediction to explicitly enforce the invariance and use data augmentation as style intervention on the RGB observation space. Our design leverages unsupervised representation learning to learn the world model with invariant causal features. Our proposed method significantly outperforms current state-of-the-art model-based and model-free reinforcement learning methods on out-of-distribution point navigation tasks on the iGibson dataset. Moreover, our proposed model excels at the sim-to-real transfer of our perception learning module. Finally, we evaluate our approach on the DeepMind control suite and enforce invariance only implicitly since depth is not available. Nevertheless, our proposed model performs on par with the state-of-the-art counterpart.

The desire to make applications and machines more intelligent and the aspiration to enable their operation without human interaction have been driving innovations in neural networks, deep learning, and other machine learning techniques. Although reinforcement learning has been primarily used in video games, recent advancements and the development of diverse and powerful reinforcement algorithms have enabled the reinforcement learning community to move from playing video games to solving complex real-life problems in autonomous systems such as self-driving cars, delivery drones, and automated robotics. Understanding the environment of an application and the algorithms' limitations plays a vital role in selecting the appropriate reinforcement learning algorithm that successfully solves the problem on hand in an efficient manner. Consequently, in this study, we identify three main environment types and classify reinforcement learning algorithms according to those environment types. Moreover, within each category, we identify relationships between algorithms. The overview of each algorithm provides insight into the algorithms' foundations and reviews similarities and differences among algorithms. This study provides a perspective on the field and helps practitioners and researchers to select the appropriate algorithm for their use case.

We introduce ASPiRe (Adaptive Skill Prior for RL), a new approach that leverages prior experience to accelerate reinforcement learning. Unlike existing methods that learn a single skill prior from a large and diverse dataset, our framework learns a library of different distinction skill priors (i.e., behavior priors) from a collection of specialized datasets, and learns how to combine them to solve a new task. This formulation allows the algorithm to acquire a set of specialized skill priors that are more reusable for downstream tasks; however, it also brings up additional challenges of how to effectively combine these unstructured sets of skill priors to form a new prior for new tasks. Specifically, it requires the agent not only to identify which skill prior(s) to use but also how to combine them (either sequentially or concurrently) to form a new prior. To achieve this goal, ASPiRe includes Adaptive Weight Module (AWM) that learns to infer an adaptive weight assignment between different skill priors and uses them to guide policy learning for downstream tasks via weighted Kullback-Leibler divergences. Our experiments demonstrate that ASPiRe can significantly accelerate the learning of new downstream tasks in the presence of multiple priors and show improvement on competitive baselines.

Humans are skillful navigators: We aptly maneuver through new places, realize when we are back at a location we have seen before, and can even conceive of shortcuts that go through parts of our environments we have never visited. Current methods in model-based reinforcement learning on the other hand struggle with generalizing about environment dynamics out of the training distribution. We argue that two principles can help bridge this gap: latent learning and parsimonious dynamics. Humans tend to think about environment dynamics in simple terms -- we reason about trajectories not in reference to what we expect to see along a path, but rather in an abstract latent space, containing information about the places' spatial coordinates. Moreover, we assume that moving around in novel parts of our environment works the same way as in parts we are familiar with. These two principles work together in tandem: it is in the latent space that the dynamics show parsimonious characteristics. We develop a model that learns such parsimonious dynamics. Using a variational objective, our model is trained to reconstruct experienced transitions in a latent space using locally linear transformations, while encouraged to invoke as few distinct transformations as possible. Using our framework, we demonstrate the utility of learning parsimonious latent dynamics models in a range of policy learning and planning tasks.

We propose a new objective for option discovery that emphasizes the computational advantage of using options in planning. In a sequential machine, the speed of planning is proportional to the number of elementary operations used to achieve a good policy. For episodic tasks, the number of elementary operations depends on the number of options composed by the policy in an episode and the number of options being considered at each decision point. To reduce the amount of computation in planning, for a given set of episodic tasks and a given number of options, our objective prefers options with which it is possible to achieve a high return by composing few options, and also prefers a smaller set of options to choose from at each decision point. We develop an algorithm that optimizes the proposed objective. In a variant of the classic four-room domain, we show that 1) a higher objective value is typically associated with fewer number of elementary planning operations used by the option-value iteration algorithm to obtain a near-optimal value function, 2) our algorithm achieves an objective value that matches it achieved by two human-designed options 3) the amount of computation used by option-value iteration with options discovered by our algorithm matches it with the human-designed options, 4) the options produced by our algorithm also make intuitive sense--they seem to move to and terminate at the entrances of rooms.

The DNA sequences that contain the same motif, a specific sequence pattern, are often bound by a particular transcriptional factor (TF) or TF combination. Biologists have shown that TF are crucial in biological processes such as alternative splicing [1], RNA degradation [2], and transcriptional regulation [3]. Different types of cells express unique combinations of TFs, which might be viewed as the basic mechanism for cell differentiation [4]. Identifying the motif from a collection of unlabeled DNA sequences that hold common regulatory or functional characteristics is a essential task for computational biology known as DNA motif discovery (DMD) (Figure 1). The input for the DMD is thousands of sequences, each containing hundreds of nucleotides. Unknown portions of these sequences contain the motif, while the remaining sequences do not.

StarCraft II (SC2) poses a grand challenge for reinforcement learning (RL), of which the main difficulties include huge state space, varying action space, and a long time horizon. In this work, we investigate a set of RL techniques for the full-length game of StarCraft II. We investigate a hierarchical RL approach, where the hierarchy involves two. One is the extracted macro-actions from experts’ demonstration trajectories to reduce the action space in an order of magnitude. The other is a hierarchical architecture of neural networks, which is modular and facilitates scale. We investigate a curriculum transfer training procedure that trains the agent from the simplest level to the hardest level. We train the agent on a single machine with 4 GPUs and 48 CPU threads. On a 64x64 map and using restrictive units, we achieve a win rate of 99% against the difficulty level-1 built-in AI. Through the curriculum transfer learning algorithm and a mixture of combat models, we achieve a 93% win rate against the most difficult non-cheating level built-in AI (level-7). In this extended version of the paper, we improve our architecture to train the agent against the most difficult cheating level AIs (level-8, level-9, and level-10). We also test our method on different maps to evaluate the extensibility of our approach. By a final 3-layer hierarchical architecture and applying significant tricks to train SC2 agents, we increase the win rate against the level-8, level-9, and level-10 to 96%, 97%, and 94%, respectively. Our codes and models are all open-sourced now at https://github.com/liuruoze/HierNet-SC2. To provide a baseline referring the AlphaStar for our work as well as the research and open-source community, we reproduce a scaled-down version of it, mini-AlphaStar (mAS). The latest version of mAS is 1.07, which can be trained using supervised learning and reinforcement learning on the raw action space which has 564 actions. It is designed to run training on a single common machine, by making the hyper-parameters adjustable and some settings simplified. We then can compare our work with mAS using the same computing resources and training time. By experiment results, we show that our method is more effective when using limited resources. The inference and training codes of mini-AlphaStar are all open-sourced at https://github.com/liuruoze/mini-AlphaStar. We hope our study could shed some light on the future research of efficient reinforcement learning on SC2 and other large-scale games.

We present Sparrow, an information-seeking dialogue agent trained to be more helpful, correct, and harmless compared to prompted language model baselines. We use reinforcement learning from human feedback to train our models with two new additions to help human raters judge agent behaviour. First, to make our agent more helpful and harmless, we break down the requirements for good dialogue into natural language rules the agent should follow, and ask raters about each rule separately. We demonstrate that this breakdown enables us to collect more targeted human judgements of agent behaviour and allows for more efficient rule-conditional reward models. Second, our agent provides evidence from sources supporting factual claims when collecting preference judgements over model statements. For factual questions, evidence provided by Sparrow supports the sampled response 78% of the time. Sparrow is preferred more often than baselines while being more resilient to adversarial probing by humans, violating our rules only 8% of the time when probed. Finally, we conduct extensive analyses showing that though our model learns to follow our rules it can exhibit distributional biases.

We present a framework to integrate tensor network (TN) methods with reinforcement learning (RL) for solving dynamical optimisation tasks. We consider the RL actor-critic method, a model-free approach for solving RL problems, and introduce TNs as the approximators for its policy and value functions. Our "actor-critic with tensor networks" (ACTeN) method is especially well suited to problems with large and factorisable state and action spaces. As an illustration of the applicability of ACTeN we solve the exponentially hard task of sampling rare trajectories in two paradigmatic stochastic models, the East model of glasses and the asymmetric simple exclusion process (ASEP), the latter being particularly challenging to other methods due to the absence of detailed balance. With substantial potential for further integration with the vast array of existing RL methods, the approach introduced here is promising both for applications in physics and to multi-agent RL problems more generally.

In edge computing, users' service profiles must be migrated in response to user mobility. Reinforcement learning (RL) frameworks have been proposed to do so. Nevertheless, these frameworks do not consider occasional server failures, which although rare, can prevent the smooth and safe functioning of edge computing users' latency sensitive applications such as autonomous driving and real-time obstacle detection, because users' computing jobs can no longer be completed. As these failures occur at a low probability, it is difficult for RL algorithms, which are inherently data-driven, to learn an optimal service migration solution for both the typical and rare event scenarios. Therefore, we introduce a rare events adaptive resilience framework FIRE, which integrates importance sampling into reinforcement learning to place backup services. We sample rare events at a rate proportional to their contribution to the value function, to learn an optimal policy. Our framework balances service migration trade-offs between delay and migration costs, with the costs of failure and the costs of backup placement and migration. We propose an importance sampling based Q-learning algorithm, and prove its boundedness and convergence to optimality. Following which we propose novel eligibility traces, linear function approximation and deep Q-learning versions of our algorithm to ensure it scales to real-world scenarios. We extend our framework to cater to users with different risk tolerances towards failure. Finally, we use trace driven experiments to show that our algorithm gives cost reductions in the event of failures.