The security of Deep Reinforcement Learning (Deep RL) algorithms deployed in real life applications are of a primary concern. In particular, the robustness of RL agents in cyber-physical systems against adversarial attacks are especially vital since the cost of a malevolent intrusions can be extremely high. Studies have shown Deep Neural Networks (DNN), which forms the core decision-making unit in most modern RL algorithms, are easily subjected to adversarial attacks. Hence, it is imperative that RL agents deployed in real-life applications have the capability to detect and mitigate adversarial attacks in an online fashion. An example of such a framework is the Meta-Learned Advantage Hierarchy (MLAH) agent that utilizes a meta-learning framework to learn policies robustly online. Since the mechanism of this framework are still not fully explored, we conducted multiple experiments to better understand the framework's capabilities and limitations. Our results shows that the MLAH agent exhibits interesting coping behaviors when subjected to different adversarial attacks to maintain a nominal reward. Additionally, the framework exhibits a hierarchical coping capability, based on the adaptability of the Master policy and sub-policies themselves. From empirical results, we also observed that as the interval of adversarial attacks increase, the MLAH agent can maintain a higher distribution of rewards, though at the cost of higher instabilities.

In complex tasks, such as those with large combinatorial action spaces, random exploration may be too inefficient to achieve meaningful learning progress. In this work, we use a curriculum of progressively growing action spaces to accelerate learning. We assume the environment is out of our control, but that the agent may set an internal curriculum by initially restricting its action space. Our approach uses off-policy reinforcement learning to estimate optimal value functions for multiple action spaces simultaneously and efficiently transfers data, value estimates, and state representations from restricted action spaces to the full task. We show the efficacy of our approach in proof-of-concept control tasks and on challenging large-scale StarCraft micromanagement tasks with large, multi-agent action spaces.

For AI community, the development of agents that react differently to different types of rewards can enable us to understand a wide spectrum of multi-agent interactions in complex real-world socioeconomic systems. Empirically, the proposed model outperforms Q-Learning and Double Q-Learning in artificial scenarios with certain reward distributions and real-world human decision making gambling tasks. Moreover, from the behavioral modeling perspective, our parametric framework can be viewed as a first step towards a unifying computational model capturing reward processing abnormalities across multiple mental conditions and user preferences in long-term recommendation systems.

Reinforcement learning has been successfully used to solve difficult tasks in complex unknown environments. However, these methods typically do not provide any safety guarantees during the learning process. This is particularly problematic, since reinforcement learning agent actively explore their environment. This prevents their use in safety-critical, real-world applications. In this paper, we present a learning-based model predictive control scheme that provides high-probability safety guarantees throughout the learning process. Based on a reliable statistical model, we construct provably accurate confidence intervals on predicted trajectories. Unlike previous approaches, we allow for input-dependent uncertainties. Based on these reliable predictions, we guarantee that trajectories satisfy safety constraints. Moreover, we use a terminal set constraint to recursively guarantee the existence of safe control actions at every iteration. We evaluate the resulting algorithm to safely explore the dynamics of an inverted pendulum and to solve a reinforcement learning task on a cart-pole system with safety constraints.

In this work we present a novel approach for transfer-guided exploration in reinforcement learning that is inspired by the human tendency to leverage experiences from similar encounters in the past while navigating a new task. Given an optimal policy in a related task-environment, we show that its bisimulation distance from the current task-environment gives a lower bound on the optimal advantage of state-action pairs in the current task-environment. Transfer-guided Exploration (ExTra) samples actions from a Softmax distribution over these lower bounds. In this way, actions with potentially higher optimum advantage are sampled more frequently. In our experiments on gridworld environments, we demonstrate that given access to an optimal policy in a related task-environment, ExTra can outperform popular domain-specific exploration strategies viz. epsilon greedy, Model-Based Interval Estimation - Exploration Based (MBIE-EB), Pursuit and Boltzmann in terms of sample complexity and rate of convergence. We further show that ExTra is robust to choices of source task and shows a graceful degradation of performance as the dissimilarity of the source task increases. We also demonstrate that ExTra, when used alongside traditional exploration algorithms, improves their rate of convergence. Thus it is capable of complimenting the efficacy of traditional exploration algorithms.

Hyperparameter tuning is an omnipresent problem in machine learning as it is an integral aspect of obtaining the state-of-the-art performance for any model. Most often, hyperparameters are optimized just by training a model on a grid of possible hyperparameter values and taking the one that performs best on a validation sample (grid search). More recently, methods have been introduced that build a so-called surrogate model that predicts the validation loss for a specific hyperparameter setting, model and dataset and then sequentially select the next hyperparameter to test, based on a heuristic function of the expected value and the uncertainty of the surrogate model called acquisition function (sequential model-based Bayesian optimization, SMBO). In this paper we model the hyperparameter optimization problem as a sequential decision problem, which hyperparameter to test next, and address it with reinforcement learning. This way our model does not have to rely on a heuristic acquisition function like SMBO, but can learn which hyperparameters to test next based on the subsequent reduction in validation loss they will eventually lead to, either because they yield good models themselves or because they allow the hyperparameter selection policy to build a better surrogate model that is able to choose better hyperparameters later on. Experiments on a large battery of 50 data sets demonstrate that our method outperforms the state-of-the-art approaches for hyperparameter learning.

Policy gradient based reinforcement learning algorithms coupled with neural networks have shown success in learning complex policies in the model free continuous action space control setting. However, explicitly parameterized policies are limited by the scope of the chosen parametric probability distribution. We show that alternatively to the likelihood based policy gradient, a related objective can be optimized through advantage weighted quantile regression. Our approach models the policy implicitly in the network, which gives the agent the freedom to approximate any distribution in each action dimension, not limiting its capabilities to the commonly used unimodal Gaussian parameterization. This broader spectrum of policies makes our algorithm suitable for problems where Gaussian policies cannot fit the optimal policy. Moreover, our results on the MuJoCo physics simulator benchmarks are comparable or superior to state-of-the-art on-policy methods.

The successful application of flexible, general learning algorithms -- such as deep reinforcement learning -- to real-world robotics applications is often limited by their poor data-efficiency. Domains with more than a single dominant task of interest encourage algorithms that share partial solutions across tasks to limit the required experiment time. We develop and investigate simple hierarchical inductive biases -- in the form of structured policies -- as a mechanism for knowledge transfer across tasks in reinforcement learning (RL). To leverage the power of these structured policies we design an RL algorithm that enables stable and fast learning. We demonstrate the success of our method both in simulated robot environments (using locomotion and manipulation domains) as well as real robot experiments, demonstrating substantially better data-efficiency than competitive baselines.

Decision making in dense traffic can be challenging for autonomous vehicles. An autonomous system only relying on predefined road priorities and considering other drivers as moving objects will cause the vehicle to freeze and fail the maneuver. Human drivers leverage the cooperation of other drivers to avoid such deadlock situations and convince others to change their behavior. Decision making algorithms must reason about the interaction with other drivers and anticipate a broad range of driver behaviors. In this work, we present a reinforcement learning approach to learn how to interact with drivers with different cooperation levels. We enhanced the performance of traditional reinforcement learning algorithms by maintaining a belief over the level of cooperation of other drivers. We show that our agent successfully learns how to navigate a dense merging scenario with less deadlocks than with online planning methods.

Machine and reinforcement learning (RL) are being applied to plan and control the behavior of autonomous systems interacting with the physical world -- examples include self-driving vehicles, distributed sensor networks, and agile robots. However, if machine learning is to be applied in these new settings, the resulting algorithms must come with the reliability, robustness, and safety guarantees that are hallmarks of the control theory literature, as failures could be catastrophic. Thus, as RL algorithms are increasingly and more aggressively deployed in safety critical settings, it is imperative that control theorists be part of the conversation. The goal of this tutorial paper is to provide a jumping off point for control theorists wishing to work on RL related problems by covering recent advances in bridging learning and control theory, and by placing these results within the appropriate historical context of the system identification and adaptive control literatures.