We present OtoWorld, an interactive environment in which agents must learn to listen in order to solve navigational tasks. The purpose of OtoWorld is to facilitate reinforcement learning research in computer audition, where agents must learn to listen to the world around them to navigate. OtoWorld is built on three open source libraries: OpenAI Gym for environment and agent interaction, PyRoomAcoustics for ray-tracing and acoustics simulation, and nussl for training deep computer audition models. OtoWorld is the audio analogue of GridWorld, a simple navigation game. OtoWorld can be easily extended to more complex environments and games. To solve one episode of OtoWorld, an agent must move towards each sounding source in the auditory scene and "turn it off". The agent receives no other input than the current sound of the room. The sources are placed randomly within the room and can vary in number. The agent receives a reward for turning off a source. We present preliminary results on the ability of agents to win at OtoWorld. OtoWorld is open-source and available.

As the applications of deep reinforcement learning (DRL) in wireless communications grow, sensitivity of DRL based wireless communication strategies against adversarial attacks has started to draw increasing attention. In order to address such sensitivity and alleviate the resulting security concerns, we in this paper consider a victim user that performs DRL-based dynamic channel access, and an attacker that executes DRLbased jamming attacks to disrupt the victim. Hence, both the victim and attacker are DRL agents and can interact with each other, retrain their models, and adapt to opponents' policies. In this setting, we initially develop an adversarial jamming attack policy that aims at minimizing the accuracy of victim's decision making on dynamic channel access. Subsequently, we devise defense strategies against such an attacker, and propose three defense strategies, namely diversified defense with proportional-integral-derivative (PID) control, diversified defense with an imitation attacker, and defense via orthogonal policies. We design these strategies to maximize the attacked victim's accuracy and evaluate their performances.

Reinforcement learning (RL) agents are often designed specifically for a particular problem and they generally have uninterpretable working processes. Statistical methods-based agent algorithms can be improved in terms of generalizability and interpretability using symbolic Artificial Intelligence (AI) tools such as logic programming. In this study, we present a model-free RL architecture that is supported with explicit relational representations of the environmental objects. For the first time, we use the PrediNet network architecture in a dynamic decision-making problem rather than image-based tasks, and Multi-Head Dot-Product Attention Network (MHDPA) as a baseline for performance comparisons. We tested two networks in two environments ---i.e., the baseline Box-World environment and our novel environment, Relational-Grid-World (RGW). With the procedurally generated RGW environment, which is complex in terms of visual perceptions and combinatorial selections, it is easy to measure the relational representation performance of the RL agents. The experiments were carried out using different configurations of the environment so that the presented module and the environment were compared with the baselines. We reached similar policy optimization performance results with the PrediNet architecture and MHDPA; additionally, we achieved to extract the propositional representation explicitly ---which makes the agent's statistical policy logic more interpretable and tractable. This flexibility in the agent's policy provides convenience for designing non-task-specific agent architectures. The main contributions of this study are two-fold ---an RL agent that can explicitly perform relational reasoning, and a new environment that measures the relational reasoning capabilities of RL agents.

In neuroscience, attention has been shown to bidirectionally interact with reinforcement learning (RL) processes. This interaction is thought to support dimensionality reduction of task representations, restricting computations to relevant features. However, it remains unclear whether these properties can translate into real algorithmic advantages for artificial agents, especially in dynamic environments. We design a model incorporating a self-attention mechanism that implements task-state representations in semantic feature-space, and test it on a battery of Atari games. To evaluate the agent's selective properties, we add a large volume of task-irrelevant features to observations. In line with neuroscience predictions, self-attention leads to increased robustness to noise compared to benchmark models. Strikingly, this self-attention mechanism is general enough, such that it can be naturally extended to implement a transient working-memory, able to solve a partially observable maze task. Lastly, we highlight the predictive quality of attended stimuli. Because we use semantic observations, we can uncover not only which features the agent elects to base decisions on, but also how it chooses to compile more complex, relational features from simpler ones. These results formally illustrate the benefits of attention in deep RL and provide evidence for the interpretability of self-attention mechanisms.

Q-learning is a model-free reinforcement learning algorithm to learn a policy telling an agent what action to take under what circumstances. It does not require a model of the environment, and it can handle problems with stochastic transitions and rewards, without requiring adaptations. For any finite Markov decision process (FMDP), Q-learning finds an optimal policy in the sense of maximizing the expected value of the total reward over any and all successive steps, starting from the current state. Q-learning can identify an optimal action-selection policy for any given FMDP, given infinite exploration time and a partly-random policy. "Q" names the function that returns the reward used to provide the reinforcement and can be said to stand for the "quality" of an action taken in a given state.

Model-based reinforcement learning (RL) is appealing because (i) it enables planning and thus more strategic exploration, and (ii) by decoupling dynamics from rewards, it enables fast transfer to new reward functions. However, learning an accurate Markov Decision Process (MDP) over high-dimensional states (e.g., raw pixels) is extremely challenging because it requires function approximation, which leads to compounding errors. Instead, to avoid compounding errors, we propose learning an abstract MDP over abstract states: low-dimensional coarse representations of the state (e.g., capturing agent position, ignoring other objects). We assume access to an abstraction function that maps the concrete states to abstract states. In our approach, we construct an abstract MDP, which grows through strategic exploration via planning. Similar to hierarchical RL approaches, the abstract actions of the abstract MDP are backed by learned subpolicies that navigate between abstract states. Our approach achieves strong results on three of the hardest Arcade Learning Environment games (Montezuma's Revenge, Pitfall!, and Private Eye), including superhuman performance on Pitfall! without demonstrations. After training on one task, we can reuse the learned abstract MDP for new reward functions, achieving higher reward in 1000x fewer samples than model-free methods trained from scratch.

Autonomous and semi-autonomous systems for safety-critical applications require rigorous testing before deployment. Due to the complexity of these systems, formal verification may be impossible and real-world testing may be dangerous during development. Therefore, simulation-based techniques have been developed that treat the system under test as a black box during testing. Safety validation tasks include finding disturbances to the system that cause it to fail (falsification), finding the most-likely failure, and estimating the probability that the system fails. Motivated by the prevalence of safety-critical artificial intelligence, this work provides a survey of state-of-the-art safety validation techniques with a focus on applied algorithms and their modifications for the safety validation problem. We present and discuss algorithms in the domains of optimization, path planning, reinforcement learning, and importance sampling. Problem decomposition techniques are presented to help scale algorithms to large state spaces, and a brief overview of safety-critical applications is given, including autonomous vehicles and aircraft collision avoidance systems. Finally, we present a survey of existing academic and commercially available safety validation tools.

The field of reinforcement learning can be split into model-based and model-free methods. Here, we unify these approaches by casting model-free policy optimisation as amortised variational inference, and model-based planning as iterative variational inference, within a `control as hybrid inference' (CHI) framework. We present an implementation of CHI which naturally mediates the balance between iterative and amortised inference. Using a didactic experiment, we demonstrate that the proposed algorithm operates in a model-based manner at the onset of learning, before converging to a model-free algorithm once sufficient data have been collected. We verify the scalability of our algorithm on a continuous control benchmark, demonstrating that it outperforms strong model-free and model-based baselines. CHI thus provides a principled framework for harnessing the sample efficiency of model-based planning while retaining the asymptotic performance of model-free policy optimisation.

Conventional simulations on multi-exit indoor evacuation focus primarily on how to determine a reasonable exit based on numerous factors in a changing environment. Results commonly include some congested and other under-utilized exits, especially with massive pedestrians. We propose a multi-exit evacuation simulation based on Deep Reinforcement Learning (DRL), referred to as the MultiExit-DRL, which involves in a Deep Neural Network (DNN) framework to facilitate state-to-action mapping. The DNN framework applies Rainbow Deep Q-Network (DQN), a DRL algorithm that integrates several advanced DQN methods, to improve data utilization and algorithm stability, and further divides the action space into eight isometric directions for possible pedestrian choices. We compare MultiExit-DRL with two conventional multi-exit evacuation simulation models in three separate scenarios: 1) varying pedestrian distribution ratios, 2) varying exit width ratios, and 3) varying open schedules for an exit. The results show that MultiExit-DRL presents great learning efficiency while reducing the total number of evacuation frames in all designed experiments. In addition, the integration of DRL allows pedestrians to explore other potential exits and helps determine optimal directions, leading to the high efficiency of exit utilization.

Deterministic Policy Gradient (DPG) removes a level of randomness from standard randomized-action Policy Gradient (PG), and demonstrates substantial empirical success for tackling complex dynamic problems involving Markov decision processes. At the same time, though, DPG loses its ability to learn in a model-free (i.e., actor-only) fashion, frequently necessitating the use of critics in order to obtain consistent estimates of the associated policy-reward gradient. In this work, we introduce Zeroth-order Deterministic Policy Gradient (ZDPG), which approximates policy-reward gradients via two-point stochastic evaluations of the $Q$-function, constructed by properly designed low-dimensional action-space perturbations. Exploiting the idea of random horizon rollouts for obtaining unbiased estimates of the $Q$-function, ZDPG lifts the dependence on critics and restores true model-free policy learning, while enjoying built-in and provable algorithmic stability. Additionally, we present new finite sample complexity bounds for ZDPG, which improve upon existing results by up to two orders of magnitude. Our findings are supported by several numerical experiments, which showcase the effectiveness of ZDPG in a practical setting, and its advantages over both PG and Baseline PG.