Safe exploration presents a major challenge in reinforcement learning (RL): when active data collection requires deploying partially trained policies, we must ensure that these policies avoid catastrophically unsafe regions, while still enabling trial and error learning. In this paper, we target the problem of safe exploration in RL by learning a conservative safety estimate of environment states through a critic, and provably upper bound the likelihood of catastrophic failures at every training iteration. We theoretically characterize the tradeoff between safety and policy improvement, show that the safety constraints are likely to be satisfied with high probability during training, derive provable convergence guarantees for our approach, which is no worse asymptotically than standard RL, and demonstrate the efficacy of the proposed approach on a suite of challenging navigation, manipulation, and locomotion tasks. Empirically, we show that the proposed approach can achieve competitive task performance while incurring significantly lower catastrophic failure rates during training than prior methods. Videos are at this url https://sites.google.com/view/ Reinforcement learning (RL) is a powerful framework for learning-based control because it can enable agents to learn to make decisions automatically through trial and error. However, in the real world, the cost of those trials - and those errors - can be quite high: an aerial robot that attempts to fly at high speed might initially crash, and then be unable to attempt further trials due to extensive physical damage. However, learning complex skills without any failures at all is likely impossible.

We introduce the $\gamma$-model, a predictive model of environment dynamics with an infinite probabilistic horizon. Replacing standard single-step models with $\gamma$-models leads to generalizations of the procedures that form the foundation of model-based control, including the model rollout and model-based value estimation. The $\gamma$-model, trained with a generative reinterpretation of temporal difference learning, is a natural continuous analogue of the successor representation and a hybrid between model-free and model-based mechanisms. Like a value function, it contains information about the long-term future; like a standard predictive model, it is independent of task reward. We instantiate the $\gamma$-model as both a generative adversarial network and normalizing flow, discuss how its training reflects an inescapable tradeoff between training-time and testing-time compounding errors, and empirically investigate its utility for prediction and control.

General value functions (GVFs) in the reinforcement learning (RL) literature are long-term predictive summaries of the outcomes of agents following specific policies in the environment. Affordances as perceived valences of action possibilities may be cast into predicted policy-relative goodness and modelled as GVFs. A systematic explication of this connection shows that GVFs and especially their deep learning embodiments (1) realize affordance prediction as a form of direct perception, (2) illuminate the fundamental connection between action and perception in affordance, and (3) offer a scalable way to learn affordances using RL methods. Through a comprehensive review of existing literature on recent successes of GVF applications in robotics, rehabilitation, industrial automation, and autonomous driving, we demonstrate that GVFs provide the right framework for learning affordances in real-world applications. In addition, we highlight a few new avenues of research opened up by the perspective of "affordance as GVF", including using GVFs for orchestrating complex behaviors.

As we deploy reinforcement learning agents to solve increasingly challenging problems, methods that allow us to inject prior knowledge about the structure of the world and effective solution strategies becomes increasingly important. In this work we consider how information and architectural constraints can be combined with ideas from the probabilistic modeling literature to learn behavior priors that capture the common movement and interaction patterns that are shared across a set of related tasks or contexts. For example the day-to day behavior of humans comprises distinctive locomotion and manipulation patterns that recur across many different situations and goals. We discuss how such behavior patterns can be captured using probabilistic trajectory models and how these can be integrated effectively into reinforcement learning schemes, e.g.\ to facilitate multi-task and transfer learning. We then extend these ideas to latent variable models and consider a formulation to learn hierarchical priors that capture different aspects of the behavior in reusable modules. We discuss how such latent variable formulations connect to related work on hierarchical reinforcement learning (HRL) and mutual information and curiosity based objectives, thereby offering an alternative perspective on existing ideas. We demonstrate the effectiveness of our framework by applying it to a range of simulated continuous control domains.

Distant supervision relation extraction (DSRE) is an efficient method to extract semantic relations on a large-scale heuristic labeling corpus. However, it usually brings in a massive noisy data. In order to alleviate this problem, many recent approaches adopt reinforcement learning (RL), which aims to select correct data autonomously before relation classification. Although these RL methods outperform conventional multi-instance learning-based methods, there are still two neglected problems: 1) the existing RL methods ignore the feedback of noisy data, 2) the reduction of training corpus exacerbates long-tail problem. In this paper, we propose a novel framework to solve the two problems mentioned above. Firstly, we design a novel reward function to obtain feedback from both correct and noisy data. In addition, we use implicit relations information to improve RL. Secondly, we propose the hierarchical memory extractor (HME), which utilizes the gating mechanism to share the semantics from correlative instances between data-rich and data-poor classes. Moreover, we define a hierarchical weighted ranking loss function to implement top-down search processing. Extensive experiments conducted on the widely used NYT dataset show significant improvement over state-of-the-art baseline methods.

AI researchers developing reinforcement learning agents could learn a lot from animals. In a decades-long venture to advance machine intelligence, the AI research community has often looked to neuroscience and behavioral science for inspiration and to better understand how intelligence is formed. But this effort has focused primarily on human intelligence, specifically that of babies and children. "This is especially true in a reinforcement learning context, where, thanks to progress in deep learning, it is now possible to bring the methods of comparative cognition directly to bear," the researchers' paper reads. "Animal cognition supplies a compendium of well-understood, nonlinguistic, intelligent behavior; it suggests experimental methods for evaluation and benchmarking; and it can guide environment and task design." DeepMind introduced some of the first forms of AI to combine deep learning and reinforcement learning, like the deep Q-network (DQN) algorithm, a system that played numerous Atari games at superhuman levels.

Detailed routing is one of the most critical steps in analog circuit design. Complete routing has become increasingly more challenging in advanced node analog circuits, making advances in efficient automatic routers ever more necessary. In this work, we propose a machine learning driven method for solving the track-assignment detailed routing problem for advanced node analog circuits. Our approach adopts an attention-based reinforcement learning (RL) policy model. Our main insight and advancement over this RL model is the use of supervision as a way to leverage solutions generated by a conventional genetic algorithm (GA). For this, our approach minimizes the Kullback-Leibler divergence loss between the output from the RL policy model and a solution distribution obtained from the genetic solver. The key advantage of this approach is that the router can learn a policy in an offline setting with supervision, while improving the run-time performance nearly 100x over the genetic solver. Moreover, the quality of the solutions our approach produces matches well with those generated by GA. We show that especially for complex problems, our supervised RL method provides good quality solution similar to conventional attention-based RL without comprising run time performance. The ability to learn from example designs and train the router to get similar solutions with orders of magnitude run-time improvement can impact the design flow dramatically, potentially enabling increased design exploration and routability-driven placement.

Decision making for autonomous driving in urban environments is challenging due to the complexity of the road structure and the uncertainty in the behavior of diverse road users. Traditional methods consist of manually designed rules as the driving policy, which require expert domain knowledge, are difficult to generalize and might give sub-optimal results as the environment gets complex. Whereas, using reinforcement learning, optimal driving policy could be learned and improved automatically through several interactions with the environment. However, current research in the field of reinforcement learning for autonomous driving is mainly focused on highway setup with little to no emphasis on urban environments. In this work, a deep reinforcement learning based decision-making approach for high-level driving behavior is proposed for urban environments in the presence of pedestrians. For this, the use of Deep Recurrent Q-Network (DRQN) is explored, a method combining state-of-the art Deep Q-Network (DQN) with a long term short term memory (LSTM) layer helping the agent gain a memory of the environment. A 3-D state representation is designed as the input combined with a well defined reward function to train the agent for learning an appropriate behavior policy in a real-world like urban simulator. The proposed method is evaluated for dense urban scenarios and compared with a rule-based approach and results show that the proposed DRQN based driving behavior decision maker outperforms the rule-based approach.

We present Revel, a partially neural reinforcement learning (RL) framework for provably safe exploration in continuous state and action spaces. A key challenge for provably safe deep RL is that repeatedly verifying neural networks within a learning loop is computationally infeasible. We address this challenge using two policy classes: a general, neurosymbolic class with approximate gradients and a more restricted class of symbolic policies that allows efficient verification. Our learning algorithm is a mirror descent over policies: in each iteration, it safely lifts a symbolic policy into the neurosymbolic space, performs safe gradient updates to the resulting policy, and projects the updated policy into the safe symbolic subset, all without requiring explicit verification of neural networks. Our empirical results show that Revel enforces safe exploration in many scenarios in which Constrained Policy Optimization does not, and that it can discover policies that outperform those learned through prior approaches to verified exploration.

Learning heuristics for combinatorial optimization problems through graph neural networks have recently shown promising results on some classic NP-hard problems. These are single-level optimization problems with only one player. Multilevel combinatorial optimization problems are their generalization, encompassing situations with multiple players taking decisions sequentially. By framing them in a multi-agent reinforcement learning setting, we devise a value-based method to learn to solve multilevel budgeted combinatorial problems involving two players in a zero-sum game over a graph. Our framework is based on a simple curriculum: if an agent knows how to estimate the value of instances with budgets up to $B$, then solving instances with budget $B+1$ can be done in polynomial time regardless of the direction of the optimization by checking the value of every possible afterstate. Thus, in a bottom-up approach, we generate datasets of heuristically solved instances with increasingly larger budgets to train our agent. We report results close to optimality on graphs up to $100$ nodes and a $185 \times$ speedup on average compared to the quickest exact solver known for the Multilevel Critical Node problem, a max-min-max trilevel problem that has been shown to be at least $\Sigma_2^p$-hard.