Vision-based reinforcement learning (RL) is a promising technique to solve control tasks involving images as the main observation. State-of-the-art RL algorithms still struggle in terms of sample efficiency, especially when using image observations. This has led to an increased attention on integrating state representation learning (SRL) techniques into the RL pipeline. Work in this field demonstrates a substantial improvement in sample efficiency among other benefits. However, to take full advantage of this paradigm, the quality of samples used for training plays a crucial role. More importantly, the diversity of these samples could affect the sample efficiency of vision-based RL, but also its generalization capability. In this work, we present an approach to improve the sample diversity. Our method enhances the exploration capability of the RL algorithms by taking advantage of the SRL setup. Our experiments show that the presented approach outperforms the baseline for all tested environments. These results are most apparent for environments where the baseline method struggles. Even in simple environments, our method stabilizes the training, reduces the reward variance and boosts sample efficiency.

Enabling humans to identify potential flaws in an agent's decision making is an important Explainable AI application. We consider identifying such flaws in a planning-based deep reinforcement learning (RL) agent for a complex real-time strategy game. In particular, the agent makes decisions via tree search using a learned model and evaluation function over interpretable states and actions. This gives the potential for humans to identify flaws at the level of reasoning steps in the tree, even if the entire reasoning process is too complex to understand. However, it is unclear whether humans will be able to identify such flaws due to the size and complexity of trees. We describe a user interface and case study, where a small group of AI experts and developers attempt to identify reasoning flaws due to inaccurate agent learning. Overall, the interface allowed the group to identify a number of significant flaws of varying types, demonstrating the promise of this approach.

Machine learning (ML) systems are rapidly increasing in size, are acquiring new capabilities, and are increasingly deployed in high-stakes settings. As with other powerful technologies, safety for ML should be a leading research priority. In response to emerging safety challenges in ML, such as those introduced by recent large-scale models, we provide a new roadmap for ML Safety and refine the technical problems that the field needs to address. We present four problems ready for research, namely withstanding hazards ("Robustness"), identifying hazards ("Monitoring"), steering ML systems ("Alignment"), and reducing risks to how ML systems are handled ("External Safety"). Throughout, we clarify each problem's motivation and provide concrete research directions.

Chekhov's gun is a dramatic principle stating that every element in a story must be necessary, and irrelevant elements should be removed. This paper presents a new natural language processing task -- Chekhov's gun recognition or (CGR) -- recognition of entities that are pivotal for the development of the plot. Though similar to classical Named Entity Recognition (NER) it has profound differences and is crucial for the tasks of narrative processing, since Chekhov's guns have a profound impact on the causal relationship in a story. The paper presents a new benchmark dataset for the CGR task that includes 5550 descriptions with one or more Chekhov's Gun in each and validates the task on two more datasets available in the natural language processing (NLP) literature. "One must never place a loaded rifle on the stage if it isn't going to go off. It's wrong to make promises you don't mean to keep."

We propose a exploration mechanism of policy in Deep Reinforcement Learning, which is exploring more when agent needs, called Add Noise to Noise (AN2N). The core idea is: when the Deep Reinforcement Learning agent is in a state of poor performance in history, it needs to explore more. So we use cumulative rewards to evaluate which past states the agents have not performed well, and use cosine distance to measure whether the current state needs to be explored more. This method shows that the exploration mechanism of the agent's policy is conducive to efficient exploration. We combining the proposed exploration mechanism AN2N with Deep Deterministic Policy Gradient (DDPG), Soft Actor-Critic (SAC) algorithms, and apply it to the field of continuous control tasks, such as halfCheetah, Hopper, and Swimmer, achieving considerable improvement in performance and convergence speed.

We present a model-free reinforcement learning algorithm to find an optimal policy for a finite-horizon Markov decision process while guaranteeing a desired lower bound on the probability of satisfying a signal temporal logic (STL) specification. We propose a method to effectively augment the MDP state space to capture the required state history and express the STL objective as a reachability objective. The planning problem can then be formulated as a finite-horizon constrained Markov decision process (CMDP). For a general finite horizon CMDP problem with unknown transition probability, we develop a reinforcement learning scheme that can leverage any model-free RL algorithm to provide an approximately optimal policy out of the general space of non-stationary randomized policies. We illustrate the effectiveness of our approach in the context of robotic motion planning for complex missions under uncertainty and performance objectives.

Several machine learning and deep learning frameworks have been proposed to solve remaining useful life estimation and failure prediction problems in recent years. Having access to the remaining useful life estimation or likelihood of failure in near future helps operators to assess the operating conditions and, therefore, provides better opportunities for sound repair and maintenance decisions. However, many operators believe remaining useful life estimation and failure prediction solutions are incomplete answers to the maintenance challenge. They argue that knowing the likelihood of failure in the future is not enough to make maintenance decisions that minimize costs and keep the operators safe. In this paper, we present a maintenance framework based on offline supervised deep reinforcement learning that instead of providing information such as likelihood of failure, suggests actions such as "continuation of the operation" or "the visitation of the repair shop" to the operators in order to maximize the overall profit. Using offline reinforcement learning makes it possible to learn the optimum maintenance policy from historical data without relying on expensive simulators. We demonstrate the application of our solution in a case study using the NASA C-MAPSS dataset.

Deep reinforcement learning (RL) algorithms can learn complex policies to optimize agent operation over time. RL algorithms have shown promising results in solving complicated problems in recent years. However, their application on real-world physical systems remains limited. Despite the advancements in RL algorithms, the industries often prefer traditional control strategies. Traditional methods are simple, computationally efficient and easy to adjust. In this paper, we first propose a new Q-learning algorithm for continuous action space, which can bridge the control and RL algorithms and bring us the best of both worlds. Our method can learn complex policies to achieve long-term goals and at the same time it can be easily adjusted to address short-term requirements without retraining. Next, we present an approximation of our algorithm which can be applied to address short-term requirements of any pre-trained RL algorithm. The case studies demonstrate that both our proposed method as well as its practical approximation can achieve short-term and long-term goals without complex reward functions.

When the sizes of the state and action spaces are large, solving MDPs can be computationally prohibitive even if the probability transition matrix is known. So in practice, a number of techniques are used to approximately solve the dynamic programming problem, including lookahead, approximate policy evaluation using an m-step return, and function approximation. In a recent paper, (Efroni et al. 2019) studied the impact of lookahead on the convergence rate of approximate dynamic programming. In this paper, we show that these convergence results change dramatically when function approximation is used in conjunction with lookout and approximate policy evaluation using an m-step return. Specifically, we show that when linear function approximation is used to represent the value function, a certain minimum amount of lookahead and multi-step return is needed for the algorithm to even converge. And when this condition is met, we characterize the finite-time performance of policies obtained using such approximate policy iteration. Our results are presented for two different procedures to compute the function approximation: linear least-squares regression and gradient descent.

The progress in deep reinforcement learning (RL) is heavily driven by the availability of challenging benchmarks used for training agents. However, benchmarks that are widely adopted by the community are not explicitly designed for evaluating specific capabilities of RL methods. While there exist environments for assessing particular open problems in RL (such as exploration, transfer learning, unsupervised environment design, or even language-assisted RL), it is generally difficult to extend these to richer, more complex environments once research goes beyond proof-of-concept results. We present MiniHack, a powerful sandbox framework for easily designing novel RL environments. MiniHack is a one-stop shop for RL experiments with environments ranging from small rooms to complex, procedurally generated worlds. By leveraging the full set of entities and environment dynamics from NetHack, one of the richest grid-based video games, MiniHack allows designing custom RL testbeds that are fast and convenient to use. With this sandbox framework, novel environments can be designed easily, either using a human-readable description language or a simple Python interface. In addition to a variety of RL tasks and baselines, MiniHack can wrap existing RL benchmarks and provide ways to seamlessly add additional complexity.