Growing concerns regarding the operational usage of AI models in the real-world has caused a surge of interest in explaining AI models' decisions to humans. Reinforcement Learning is not an exception in this regard. In this work, we propose a method for offering local explanations on risk in reinforcement learning. Our method only requires a log of previous interactions between the agent and the environment to create a state-transition model. It is designed to work on RL environments with either continuous or discrete state and action spaces. After creating the model, actions of any agent can be explained in terms of the features most influential in increasing or decreasing risk or any other desirable objective function in the locality of the agent. Through experiments, we demonstrate the effectiveness of the proposed method in providing such explanations.

Although distributional reinforcement learning The theoretical validity of QR-DQN [Dabney et al., (DRL) has been widely examined in the past few 2018b], IQN [Dabney et al., 2018a] and FQF [Yang et al., years, there are two open questions people are still 2019] heavily depends on a prerequisite that the approximated trying to address. One is how to ensure the validity quantile curve is non-decreasing. Unfortunately, since of the learned quantile function, the other is how to no global constraint is imposed when simultaneously estimating efficiently utilize the distribution information. This the quantile values at multiple locations, the monotonicity paper attempts to provide some new perspectives to can not be ensured using their network designs. At encourage the future in-depth studies in these two early training stage, the crossing issue is even more severe fields. We first propose a non-decreasing quantile given limited training samples. Some existing studies try to function network (NDQFN) to guarantee the monotonicity solve this problem [Zhou et al., 2020; Tang Nguyen et al., of the obtained quantile estimates and then 2020]. However, their main architecture is built on a set of design a general exploration framework called distributional fixed quantile locations and not applicable to quantile value prediction error (DPE) for DRL which based algorithms such as IQN and FQF.

There is considerable interest in designing meta-reinforcement learning (meta-RL) algorithms, which enable autonomous agents to adapt new tasks from small amount of experience. In meta-RL, the specification (such as reward function) of current task is hidden from the agent. In addition, states are hidden within each task owing to sensor noise or limitations in realistic environments. Therefore, the meta-RL agent faces the challenge of specifying both the hidden task and states based on small amount of experience. To address this, we propose estimating disentangled belief about task and states, leveraging an inductive bias that the task and states can be regarded as global and local features of each task. Specifically, we train a hierarchical state-space model (HSSM) parameterized by deep neural networks as an environment model, whose global and local latent variables correspond to task and states, respectively. Because the HSSM does not allow analytical computation of posterior distribution, i.e., belief, we employ amortized inference to approximate it. After the belief is obtained, we can augment observations of a model-free policy with the belief to efficiently train the policy. Moreover, because task and state information are factorized and interpretable, the downstream policy training is facilitated compared with the prior methods that did not consider the hierarchical nature. Empirical validations on a GridWorld environment confirm that the HSSM can separate the hidden task and states information. Then, we compare the meta-RL agent with the HSSM to prior meta-RL methods in MuJoCo environments, and confirm that our agent requires less training data and reaches higher final performance.

Model based reinforcement learning (MBRL) uses an imperfect model of the world to imagine trajectories of future states and plan the best actions to maximize a reward function. These trajectories are imperfect and MBRL attempts to overcome this by relying on model predictive control (MPC) to continuously re-imagine trajectories from scratch. Such re-generation of imagined trajectories carries the major computational cost and increasing complexity in tasks with longer receding horizon. This paper aims to investigate how far in the future the imagined trajectories can be relied upon while still maintaining acceptable reward. Firstly, an error analysis is presented for systematic skipping recalculations for varying number of consecutive steps.% in several challenging benchmark control tasks. Secondly, we propose two methods offering when to trust and act upon imagined trajectories, looking at recent errors with respect to expectations, or comparing the confidence in an action imagined against its execution. Thirdly, we evaluate the effects of acting upon imagination while training the model of the world. Results show that acting upon imagination can reduce calculations by at least 20% and up to 80%, depending on the environment, while retaining acceptable reward.

Algorithmic Information Theory has inspired intractable constructions of general intelligence (AGI), and undiscovered tractable approximations are likely feasible. Reinforcement Learning (RL), the dominant paradigm by which an agent might learn to solve arbitrary solvable problems, gives an agent a dangerous incentive: to gain arbitrary "power" in order to intervene in the provision of their own reward. We review the arguments that generally intelligent algorithmic-information-theoretic reinforcement learners such as Hutter's (2005) AIXI would seek arbitrary power, including over us. Then, using an information-theoretic exploration schedule, and a setup inspired by causal influence theory, we present a variant of AIXI which learns to not seek arbitrary power; we call it "unambitious". We show that our agent learns to accrue reward at least as well as a human mentor, while relying on that mentor with diminishing probability. And given a formal assumption that we probe empirically, we show that eventually, the agent's world-model incorporates the following true fact: intervening in the "outside world" will have no effect on reward acquisition; hence, it has no incentive to shape the outside world.

AlphaZero has achieved impressive performance in deep reinforcement learning by utilizing an architecture that combines search and training of a neural network in self-play. Many researchers are looking for ways to reproduce and improve results for other games/tasks. However, the architecture is designed to learn from scratch, tabula rasa, accepting a cold-start problem in self-play. Recently, a warm-start enhancement method for Monte Carlo Tree Search was proposed to improve the self-play starting phase. It employs a fixed parameter $I^\prime$ to control the warm-start length. Improved performance was reported in small board games. In this paper we present results with an adaptive switch method. Experiments show that our approach works better than the fixed $I^\prime$, especially for "deep," tactical, games (Othello and Connect Four). We conjecture that the adaptive value for $I^\prime$ is also influenced by the size of the game, and that on average $I^\prime$ will increase with game size. We conclude that AlphaZero-like deep reinforcement learning benefits from adaptive rollout based warm-start, as Rapid Action Value Estimate did for rollout-based reinforcement learning 15 years ago.

Biopharmaceutical manufacturing is a rapidly growing industry with impact in virtually all branches of medicine. Biomanufacturing processes require close monitoring and control, in the presence of complex bioprocess dynamics with many interdependent factors, as well as extremely limited data due to the high cost and long duration of experiments. We develop a novel model-based reinforcement learning framework that can achieve human-level control in low-data environments. The model uses a probabilistic knowledge graph to capture causal interdependencies between factors in the underlying stochastic decision process, leveraging information from existing kinetic models from different unit operations while incorporating real-world experimental data. We then present a computationally efficient, provably convergent stochastic gradient method for policy optimization. Validation is conducted on a realistic application with a multi-dimensional, continuous state variable.

In this paper, we develop a safe decision-making method for self-driving cars in a multi-lane, single-agent setting. The proposed approach utilizes deep reinforcement learning (RL) to achieve a high-level policy for safe tactical decision-making. We address two major challenges that arise solely in autonomous navigation. First, the proposed algorithm ensures that collisions never happen, and therefore accelerate the learning process. Second, the proposed algorithm takes into account the unobservable states in the environment. These states appear mainly due to the unpredictable behavior of other agents, such as cars, and pedestrians, and make the Markov Decision Process (MDP) problematic when dealing with autonomous navigation. Simulations from a well-known self-driving car simulator demonstrate the applicability of the proposed method

In Goal-oriented Reinforcement learning, relabeling the raw goals in past experience to provide agents with hindsight ability is a major solution to the reward sparsity problem. In this paper, to enhance the diversity of relabeled goals, we develop FGI (Foresight Goal Inference), a new relabeling strategy that relabels the goals by looking into the future with a learned dynamics model. Besides, to improve sample efficiency, we propose to use the dynamics model to generate simulated trajectories for policy training. By integrating these two improvements, we introduce the MapGo framework (Model-Assisted Policy Optimization for Goal-oriented tasks). In our experiments, we first show the effectiveness of the FGI strategy compared with the hindsight one, and then show that the MapGo framework achieves higher sample efficiency when compared to model-free baselines on a set of complicated tasks.

On the As one of the solutions to the Dec-POMDP problem, the value other hand, in order to extract helpful information from the state of decomposition method has achieved good results recently. However, the complex environment, some work[12, 19] promotes the neural most value decomposition methods require the global state network to learn useful state information by adding auxiliary tasks during training, but this is not feasible in some scenarios where mainly to predict the state of the next moment. Intuitively, the the global state cannot be obtained. Therefore, we propose a novel problem with these studies is in that they cannot be implemented value decomposition framework, named State Inference for value for tasks that cannot obtain real state information. DEcomposition (SIDE), which eliminates the need to know the true As a notorious problem in MAS, Dec-POMDP[25] describes some state by simultaneously seeking solutions to the two problems of collaboration problems.