Deep learning has become an increasingly common technique for various control problems, such as robotic arm manipulation, robot navigation, and autonomous vehicles. However, the downside of using deep neural networks to learn control policies is their opaque nature and the difficulties of validating their safety. As the networks used to obtain state-of-the-art results become increasingly deep and complex, the rules they have learned and how they operate become more challenging to understand. This presents an issue, since in safety-critical applications the safety of the control policy must be ensured to a high confidence level. In this paper, we propose an automated black box testing framework based on adversarial reinforcement learning. The technique uses an adversarial agent, whose goal is to degrade the performance of the target model under test. We test the approach on an autonomous vehicle problem, by training an adversarial reinforcement learning agent, which aims to cause a deep neural network-driven autonomous vehicle to collide. Two neural networks trained for autonomous driving are compared, and the results from the testing are used to compare the robustness of their learned control policies. We show that the proposed framework is able to find weaknesses in both control policies that were not evident during online testing and therefore, demonstrate a significant benefit over manual testing methods.
The Knowledge Gradient (KG) policy was originally proposed for online ranking and selection problems but has recently been adapted for use in online decision making in general and multi-armed bandit problems (MABs) in particular. We study its use in a class of exponential family MABs and identify weaknesses, including a propensity to take actions which are dominated with respect to both exploitation and exploration. We propose variants of KG which avoid such errors. These new policies include an index heuristic which deploys a KG approach to develop an approximation to the Gittins index. A num…
In the context of learning deterministic policies in continuous domains, we revisit an approach, which was first proposed in Continuous Actor Critic Learning Automaton (CACLA) and later extended in Neural Fitted Actor Critic (NFAC). This approach is based on a policy update different from that of deterministic policy gradient (DPG). Previous work has observed its excellent performance empirically, but a theoretical justification is lacking. To fill this gap, we provide a theoretical explanation to motivate this unorthodox policy update by relating it to another update and making explicit the o…
Despite the close connection between exploration and sample efficiency, most state of the art reinforcement learning algorithms include no considerations for exploration beyond maximizing the entropy of the policy. In this work we address this seeming missed opportunity. We observe that the most common formulation of directed exploration in deep RL, known as bonus-based exploration (BBE), suffers from bias and slow coverage in the few-sample regime. This causes BBE to be actively detrimental to policy learning in many control tasks. We show that by decoupling the task policy from the explorati…
Advances in computing resources have resulted in the increasing complexity of cyber-physical systems (CPS). As the complexity of CPS evolved, the focus has shifted from traditional control methods to deep reinforcement learning-based (DRL) methods for control of these systems. This is due to the difficulty of obtaining accurate models of complex CPS for traditional control. However, to securely deploy DRL in production, it is essential to examine the weaknesses of DRL-based controllers (policies) towards malicious attacks from all angles. In this work, we investigate targeted attacks in the ac…
Carnegie Mellon University（卡内基梅隆大学）；University of Texas - Austin（德克萨斯大学奥斯汀分校）
In multi-server queueing systems where there is no central queue holding all incoming jobs, job dispatching policies are used to assign incoming jobs to the queue at one of the servers. Classic job dispatching policies such as join-the-shortest-queue and shortest expected delay assume that the service rates and queue lengths of the servers are known to the dispatcher. In this work, we tackle the problem of job dispatching without the knowledge of service rates and queue lengths, where the dispatcher can only obtain noisy estimates of the service rates by observing job departures. This problem …
This paper proposes new, end-to-end deep reinforcement learning algorithms for learning two-player zero-sum Markov games. Different from prior efforts on training agents to beat a fixed set of opponents, our objective is to find the Nash equilibrium policies that are free from exploitation by even the adversarial opponents. We propose (a) Nash-DQN algorithm, which integrates the deep learning techniques from single DQN into the classic Nash Q-learning algorithm for solving tabular Markov games; (b) Nash-DQN-Exploiter algorithm, which additionally adopts an exploiter to guide the exploration of…
This work theoretically studies a ubiquitous reinforcement learning policy for controlling the canonical model of continuous-time stochastic linear-quadratic systems. We show that randomized certainty equivalent policy addresses the exploration-exploitation dilemma in linear control systems that evolve according to unknown stochastic differential equations and their operating cost is quadratic. More precisely, we establish square-root of time regret bounds, indicating that randomized certainty equivalent policy learns optimal control actions fast from a single state trajectory. Further, linear…
This paper investigates a class of attacks targeting the confidentiality aspect of security in Deep Reinforcement Learning (DRL) policies. Recent research have established the vulnerability of supervised machine learning models (e.g., classifiers) to model extraction attacks. Such attacks leverage the loosely-restricted ability of the attacker to iteratively query the model for labels, thereby allowing for the forging of a labeled dataset which can be used to train a replica of the original model. In this work, we demonstrate the feasibility of exploiting imitation learning techniques in launc…
Stanford University（斯坦福大学）；Toyota Research Institute（丰田研究院）
Autonomous driving has achieved significant progress in recent years, but autonomous cars are still unable to tackle high-risk situations where a potential accident is likely. In such near-accident scenarios, even a minor change in the vehicle's actions may result in drastically different consequences. To avoid unsafe actions in near-accident scenarios, we need to fully explore the environment. However, reinforcement learning (RL) and imitation learning (IL), two widely-used policy learning methods, cannot model rapid phase transitions and are not scalable to fully cover all the states. To add…
We propose a reinforcement learning agent to solve hard exploration games by learning a range of directed exploratory policies. We construct an episodic memory-based intrinsic reward using k-nearest neighbors over the agent's recent experience to train the directed exploratory policies, thereby encouraging the agent to repeatedly revisit all states in its environment. A self-supervised inverse dynamics model is used to train the embeddings of the nearest neighbour lookup, biasing the novelty signal towards what the agent can control. We employ the framework of Universal Value Function Approxim…
Off-policy evaluation (OPE) is the problem of evaluating new policies using historical data obtained from a different policy. In the recent OPE context, most studies have focused on single-player cases, and not on multi-player cases. In this study, we propose OPE estimators constructed by the doubly robust and double reinforcement learning estimators in two-player zero-sum Markov games. The proposed estimators project exploitability that is often used as a metric for determining how close a policy profile (i.e., a tuple of policies) is to a Nash equilibrium in two-player zero-sum games. We pro…
The purpose of this paper is to provide further understanding into the structure of the sequential allocation ("stochastic multi-armed bandit", or MAB) problem by establishing probability one finite horizon bounds and convergence rates for the sample (or "pseudo") regret associated with two simple classes of allocation policies $π$. For any slowly increasing function $g$, subject to mild regularity constraints, we construct two policies (the $g$-Forcing, and the $g$-Inflated Sample Mean) that achieve a measure of regret of order $ O(g(n))$ almost surely as $n \to \infty$, bound from above and …
Transfer learning significantly accelerates the reinforcement learning process by exploiting relevant knowledge from previous experiences. The problem of optimally selecting source policies during the learning process is of great importance yet challenging. There has been little theoretical analysis of this problem. In this paper, we develop an optimal online method to select source policies for reinforcement learning. This method formulates online source policy selection as a multi-armed bandit problem and augments Q-learning with policy reuse. We provide theoretical guarantees of the optimal…
This paper presents a framework for efficiently learning feature selection policies which use less features to reach a high classification precision on large unstructured data. It uses a Deep Convolutional Autoencoder (DCAE) for learning compact feature spaces, in combination with recently-proposed Reinforcement Learning (RL) algorithms as Double DQN and Retrace.
We consider online sequential decision problems where an agent must balance exploration and exploitation. We derive a set of Bayesian `optimistic' policies which, in the stochastic multi-armed bandit case, includes the Thompson sampling policy. We provide a new analysis showing that any algorithm producing policies in the optimistic set enjoys $\tilde O(\sqrt{AT})$ Bayesian regret for a problem with $A$ actions after $T$ rounds. We extend the regret analysis for optimistic policies to bilinear saddle-point problems which include zero-sum matrix games and constrained bandits as special cases. I…
Vector Institute（向量学院）；University of Toronto（多伦多大学）；Ryerson University（瑞尔森大学）
Reinforcement learning (RL) methods usually treat reward functions as black boxes. As such, these methods must extensively interact with the environment in order to discover rewards and optimal policies. In most RL applications, however, users have to program the reward function and, hence, there is the opportunity to make the reward function visible -- to show the reward function's code to the RL agent so it can exploit the function's internal structure to learn optimal policies in a more sample efficient manner. In this paper, we show how to accomplish this idea in two steps. First, we propo…
Queensland University of Technology (QUT)（昆士兰科技大学（QUT））
Emerging object-based SLAM algorithms can build a graph representation of an environment comprising nodes for robot poses and object landmarks. However, while this map will contain static objects such as furniture or appliances, many moveable objects (e.g. the car keys, the glasses, or a magazine), are not suitable as landmarks and will not be part of the map due to their non-static nature. We show that Graph Convolutional Networks can learn navigation policies to find such unmapped objects by learning to exploit the hidden probabilistic model that governs where these objects appear in the env…
Stanford University（斯坦福大学）；KTH Royal Institute of Technology（KTH皇家理工学院）；Aalto University（阿尔托大学）
We present a data-efficient framework for solving sequential decision-making problems which exploits the combination of reinforcement learning (RL) and latent variable generative models. The framework, called GenRL, trains deep policies by introducing an action latent variable such that the feed-forward policy search can be divided into two parts: (i) training a sub-policy that outputs a distribution over the action latent variable given a state of the system, and (ii) unsupervised training of a generative model that outputs a sequence of motor actions conditioned on the latent action variable…
We consider a finite-horizon multi-armed bandit (MAB) problem in a Bayesian setting, for which we propose an information relaxation sampling framework. With this framework, we define an intuitive family of control policies that include Thompson sampling (TS) and the Bayesian optimal policy as endpoints. Analogous to TS, which, at each decision epoch pulls an arm that is best with respect to the randomly sampled parameters, our algorithms sample entire future reward realizations and take the corresponding best action. However, this is done in the presence of "penalties" that seek to compensate …
In reinforcement learning (RL), an agent must explore an initially unknown environment in order to learn a desired behaviour. When RL agents are deployed in real world environments, safety is of primary concern. Constrained Markov decision processes (CMDPs) can provide long-term safety constraints; however, the agent may violate the constraints in an effort to explore its environment. This paper proposes a model-based RL algorithm called Explicit Explore, Exploit, or Escape ($E^{4}$), which extends the Explicit Explore or Exploit ($E^{3}$) algorithm to a robust CMDP setting. $E^4$ explicitly s…
University of California, Berkeley（加州大学伯克利分校）；Technical University of Munich（慕尼黑工业大学）
Self-play reinforcement learning has achieved state-of-the-art, and often superhuman, performance in a variety of zero-sum games. Yet prior work has found that policies that are highly capable against regular opponents can fail catastrophically against adversarial policies: an opponent trained explicitly against the victim. Prior defenses using adversarial training were able to make the victim robust to a specific adversary, but the victim remained vulnerable to new ones. We conjecture this limitation was due to insufficient diversity of adversaries seen during training. We analyze a defense u…
Automaton Distillation: Neuro-Symbolic Transfer Learning for Deep Reinforcement Learning
University of Texas at San Antonio（德克萨斯大学圣安东尼奥分校）；University of Central Florida（中佛罗里达大学）；Air Force Research Laboratory（空军研究实验室）；Florida International University（佛罗里达国际大学）；University of Colorado（科罗拉多大学）
Indian Institute of Technology Bombay（印度理工学院孟买分校）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Indian Institute of Technology Kanpur（印度理工学院坎普尔分校）
To overcome the curses of dimensionality and modeling of Dynamic Programming (DP) methods to solve Markov Decision Process (MDP) problems, Reinforcement Learning (RL) methods are adopted in practice. Contrary to traditional RL algorithms which do not consider the structural properties of the optimal policy, we propose a structure-aware learning algorithm to exploit the ordered multi-threshold structure of the optimal policy, if any. We prove the asymptotic convergence of the proposed algorithm to the optimal policy. Due to the reduction in the policy space, the proposed algorithm provides rema…
Nankai University（南开大学）；Microsoft Research Asia（微软亚洲研究院）
A fundamental issue in reinforcement learning algorithms is the balance between exploration of the environment and exploitation of information already obtained by the agent. Especially, exploration has played a critical role for both efficiency and efficacy of the learning process. However, Existing works for exploration involve task-agnostic design, that is performing well in one environment, but be ill-suited to another. To the purpose of learning an effective and efficient exploration policy in an automated manner. We formalized a feasible metric for measuring the utility of exploration bas…
We consider a new form of reinforcement learning (RL) that is based on opportunities to directly learn the optimal control policy and a general Markov decision process (MDP) framework devised to support these opportunities. Derivations of general classes of our control-based RL methods are presented, together with forms of exploration and exploitation in learning and applying the optimal control policy over time. Our general MDP framework extends the classical Bellman operator and optimality criteria by generalizing the definition and scope of a policy for any given state. We establish the con…
Training visual control policies from scratch on a new robot typically requires generating large amounts of robot-specific data. How might we leverage data previously collected on another robot to reduce or even completely remove this need for robot-specific data? We propose a "robot-aware control" paradigm that achieves this by exploiting readily available knowledge about the robot. We then instantiate this in a robot-aware model-based RL policy by training modular dynamics models that couple a transferable, robot-aware world dynamics module with a robot-specific, potentially analytical, robo…
Istituto Italiano di Tecnologia（意大利技术研究院）；Università di Genova（热那亚大学）；Bosch Center for Artificial Intelligence（博世人工智能中心）
A common strategy to deal with the expensive reinforcement learning (RL) of complex tasks is to decompose them into a collection of subtasks that are usually simpler to learn as well as reusable for new problems. However, when a robot learns the policies for these subtasks, common approaches treat every policy learning process separately. Therefore, all these individual (composable) policies need to be learned before tackling the learning process of the complex task through policies composition. Moreover, such composition of individual policies is usually performed sequentially, which is not s…
The current dominant paradigm in sensorimotor control, whether imitation or reinforcement learning, is to train policies directly in raw action spaces such as torque, joint angle, or end-effector position. This forces the agent to make decisions individually at each timestep in training, and hence, limits the scalability to continuous, high-dimensional, and long-horizon tasks. In contrast, research in classical robotics has, for a long time, exploited dynamical systems as a policy representation to learn robot behaviors via demonstrations. These techniques, however, lack the flexibility and ge…
Carnegie Mellon University（卡内基梅隆大学）；Facebook AI Research（脸书人工智能研究院）
A key challenge in reinforcement learning (RL) is environment generalization: a policy trained to solve a task in one environment often fails to solve the same task in a slightly different test environment. A common approach to improve inter-environment transfer is to learn policies that are invariant to the distribution of testing environments. However, we argue that instead of being invariant, the policy should identify the specific nuances of an environment and exploit them to achieve better performance. In this work, we propose the 'Environment-Probing' Interaction (EPI) policy, a policy t…
Robot control policies for temporally extended and sequenced tasks are often characterized by discontinuous switches between different local dynamics. These change-points are often exploited in hierarchical motion planning to build approximate models and to facilitate the design of local, region-specific controllers. However, it becomes combinatorially challenging to implement such a pipeline for complex temporally extended tasks, especially when the sub-controllers work on different information streams, time scales and action spaces. In this paper, we introduce a method that can compose diver…
We present a new model-based algorithm for reinforcement learning (RL) which consists of explicit exploration and exploitation phases, and is applicable in large or infinite state spaces. The algorithm maintains a set of dynamics models consistent with current experience and explores by finding policies which induce high disagreement between their state predictions. It then exploits using the refined set of models or experience gathered during exploration. We show that under realizability and optimal planning assumptions, our algorithm provably finds a near-optimal policy with a number of samp…
SUB-PLAY: Adversarial Policies against Partially Observed Multi-Agent Reinforcement Learning Systems
Zhejiang University（浙江大学）；Laboratory for Big Data and Decision（大数据与决策实验室）；Chinese Aeronautical Establishment（中国航空研究院）；Institute of Computer Application, China Academy of Engineering Physics（中国工程物理研究院计算机应用研究所）
This paper considers a multi-agent Markov Decision Process (MDP), where there are $n$ agents and each agent $i$ is associated with a state $s_i$ and action $a_i$ taking values from a finite set. Though the global state space size and action space size are exponential in $n$, we impose local dependence structures and focus on local policies that only depend on local states, and we propose a method that finds nearly optimal local policies in polynomial time (in $n$) when the dependence structure is a one directional tree. The algorithm builds on approximated reward functions which are evaluated …
The University of Sydney Business School（悉尼大学商学院）；University of Southampton（南安普顿大学）
In budget-limited multi-armed bandit (MAB) problems, the learner's actions are costly and constrained by a fixed budget. Consequently, an optimal exploitation policy may not be to pull the optimal arm repeatedly, as is the case in other variants of MAB, but rather to pull the sequence of different arms that maximises the agent's total reward within the budget. This difference from existing MABs means that new approaches to maximising the total reward are required. Given this, we develop two pulling policies, namely: (i) KUBE; and (ii) fractional KUBE. Whereas the former provides better perform…
Binghamton University（宾汉姆顿大学）；Yantai University（烟台大学）；Assured Information Security（Assured Information Security（Assured Information Security公司））
We study an important yet under-addressed problem of quickly and safely improving policies in online reinforcement learning domains. As its solution, we propose a novel exploration strategy - diverse exploration (DE), which learns and deploys a diverse set of safe policies to explore the environment. We provide DE theory explaining why diversity in behavior policies enables effective exploration without sacrificing exploitation. Our empirical study shows that an online policy improvement algorithm framework implementing the DE strategy can achieve both fast policy improvement and safe online p…
Australian National University（澳大利亚国立大学）；Data61, CSIRO（联邦科学与工业研究组织Data61）；Deakin University（迪肯大学）；Microsoft Research Asia（微软亚洲研究院）
We consider the problem of personalised news recommendation where each user consumes news in a sequential fashion. Existing personalised news recommendation methods focus on exploiting user interests and ignores exploration in recommendation, which leads to biased feedback loops and hurt recommendation quality in the long term. We build on contextual bandits recommendation strategies which naturally address the exploitation-exploration trade-off. The main challenges are the computational efficiency for exploring the large-scale item space and utilising the deep representations with uncertainty…
Reinforcement learning with sparse rewards is challenging because an agent can rarely obtain non-zero rewards and hence, gradient-based optimization of parameterized policies can be incremental and slow. Recent work demonstrated that using a memory buffer of previous successful trajectories can result in more effective policies. However, existing methods may overly exploit past successful experiences, which can encourage the agent to adopt sub-optimal and myopic behaviors. In this work, instead of focusing on good experiences with limited diversity, we propose to learn a trajectory-conditioned…
Recent applications of deep learning to navigation have generated end-to-end navigation solutions whereby visual sensor input is mapped to control signals or to motion primitives. The resulting visual navigation strategies work very well at collision avoidance and have performance that matches traditional reactive navigation algorithms while operating in real-time. It is accepted that these solutions cannot provide the same level of performance as a global planner. However, it is less clear how such end-to-end systems should be integrated into a full navigation pipeline. We evaluate the typica…
Interpretable and Explainable Logical Policies via Neurally Guided Symbolic Abstraction
Technical University of Darmstadt（达姆施塔特工业大学）；National Research Center for Applied Cybersecurity（国家应用网络安全研究中心）；Eindhoven University of Technology（埃因霍温理工大学）；Hessian Center for AI (hessian.AI)（黑森人工智能中心）；German Research Center for AI (DFKI)（德国人工智能研究中心）
Humans decompose novel complex tasks into simpler ones to exploit previously learned skills. Analogously, hierarchical reinforcement learning seeks to leverage lower-level policies for simple tasks to solve complex ones. However, because each lower-level policy induces a different distribution of states, transitioning from one lower-level policy to another may fail due to an unexpected starting state. We introduce transition policies that smoothly connect lower-level policies by producing a distribution of states and actions that matches what is expected by the next policy. Training transition…
In this paper, we present exploitability descent, a new algorithm to compute approximate equilibria in two-player zero-sum extensive-form games with imperfect information, by direct policy optimization against worst-case opponents. We prove that when following this optimization, the exploitability of a player's strategy converges asymptotically to zero, and hence when both players employ this optimization, the joint policies converge to a Nash equilibrium. Unlike fictitious play (XFP) and counterfactual regret minimization (CFR), our convergence result pertains to the policies being optimized …
We analyze the problem of using Explore-Exploit techniques to improve precision in multi-result ranking systems such as web search, query autocompletion and news recommendation. Adopting an exploration policy directly online, without understanding its impact on the production system, may have unwanted consequences - the system may sustain large losses, create user dissatisfaction, or collect exploration data which does not help improve ranking quality. An offline framework is thus necessary to let us decide what policy and how we should apply in a production environment to ensure positive outc…
Pacific Northwest National Laboratory（太平洋西北国家实验室）；The City College of New York（纽约市立学院）；Grove School of Engineering（格罗夫工程学院）；City University of New York（纽约城市大学）
Single-agent reinforcement learning algorithms in a multi-agent environment are inadequate for fostering cooperation. If intelligent agents are to interact and work together to solve complex problems, methods that counter non-cooperative behavior are needed to facilitate the training of multiple agents. This is the goal of cooperative AI. Recent research in adversarial machine learning, however, shows that models (e.g., image classifiers) can be easily deceived into making inferior decisions. Meanwhile, an important line of research in cooperative AI has focused on introducing algorithmic impr…
École Polytechnique Fédérale de Lausanne（洛桑联邦理工学院）
We address the problem of designing stabilizing control policies for nonlinear systems in discrete-time, while minimizing an arbitrary cost function. When the system is linear and the cost is convex, the System Level Synthesis (SLS) approach offers an effective solution based on convex programming. Beyond this case, a globally optimal solution cannot be found in a tractable way, in general. In this paper, we develop a parametrization of all and only the control policies stabilizing a given time-varying nonlinear system in terms of the combined effect of 1) a strongly stabilizing base controlle…
In this work, we propose a framework to learn feedback control policies with guarantees on closed-loop generalization and adversarial robustness. These policies are learned directly from expert demonstrations, contained in a dataset of state-control input pairs, without any prior knowledge of the task and system model. We use a Lipschitz-constrained loss minimization scheme to learn feedback policies with certified closed-loop robustness, wherein the Lipschitz constraint serves as a mechanism to tune the generalization performance and robustness to adversarial disturbances. Our analysis exploi…
Ubiquant Investment（九坤投资）；University of Chicago（芝加哥大学）；Booth School of Business, University of Chicago（芝加哥大学布斯商学院）；Carnegie Mellon University（卡内基梅隆大学）；Harvard University（哈佛大学）；FLAIR, University of Oxford（牛津大学未来智能研究院）
