In this paper, we investigate the problem of controlling probabilistic Boolean control networks (PBCNs) to achieve reachability with maximum probability in the finite time horizon. We address three questions: 1) finding control policies that achieve reachability with maximum probability under fixed, and particularly, varied finite time horizon, 2) leveraging prior knowledge to solve question 1) with faster convergence speed in scenarios where time is a variable framework, and 3) proposing an enhanced Q-learning (QL) method to efficiently address the aforementioned questions for large-scale PBCNs. For question 1), we demonstrate the applicability of QL method on the finite-time reachability problem. For question 2), considering the possibility of varied time frames, we incorporate transfer learning (TL) technique to leverage prior knowledge and enhance convergence speed. For question 3), an enhanced model-free QL approach that improves upon the traditional QL algorithm by introducing memory-efficient modifications to address these issues in large-scale PBCNs effectively. Finally, we apply the proposed method to two examples: a small-scale PBCN and a large-scale PBCN, demonstrating the effectiveness of our approach.
This paper investigates the problem of designing control policies that satisfy high-level specifications described by signal temporal logic (STL) in unknown, stochastic environments. While many existing works concentrate on optimizing the spatial robustness of a system, our work takes a step further by also considering temporal robustness as a critical metric to quantify the tolerance of time uncertainty in STL. To this end, we formulate two relevant control objectives to enhance the temporal robustness of the synthesized policies. The first objective is to maximize the probability of being temporally robust for a given threshold. The second objective is to maximize the worst-case spatial robustness value within a bounded time shift. We use reinforcement learning to solve both control synthesis problems for unknown systems. Specifically, we approximate both control objectives in a way that enables us to apply the standard Q-learning algorithm. Theoretical bounds in terms of the approximations are also derived. We present case studies to demonstrate the feasibility of our approach.
Reinforcement learning (RL) methods with a high replay ratio (RR) and regularization have gained interest due to their superior sample efficiency. However, these methods have mainly been developed for dense-reward tasks. In this paper, we aim to extend these RL methods to sparse-reward goal-conditioned tasks. We use Randomized Ensemble Double Q-learning (REDQ) (Chen et al., 2021), an RL method with a high RR and regularization. To apply REDQ to sparse-reward goal-conditioned tasks, we make the following modifications to it: (i) using hindsight experience replay and (ii) bounding target Q-values. We evaluate REDQ with these modifications on 12 sparse-reward goal-conditioned tasks of Robotics (Plappert et al., 2018), and show that it achieves about $2 \times$ better sample efficiency than previous state-of-the-art (SoTA) RL methods. Furthermore, we reconsider the necessity of specific components of REDQ and simplify it by removing unnecessary ones. The simplified REDQ with our modifications achieves $\sim 8 \times$ better sample efficiency than the SoTA methods in 4 Fetch tasks of Robotics.
Informational parsimony provides a useful inductive bias for learning representations that achieve better generalization by being robust to noise and spurious correlations. We propose \textit{information gating} as a way to learn parsimonious representations that identify the minimal information required for a task. When gating information, we can learn to reveal as little information as possible so that a task remains solvable, or hide as little information as possible so that a task becomes unsolvable. We gate information using a differentiable parameterization of the signal-to-noise ratio, which can be applied to arbitrary values in a network, e.g., erasing pixels at the input layer or activations in some intermediate layer. When gating at the input layer, our models learn which visual cues matter for a given task. When gating intermediate layers, our models learn which activations are needed for subsequent stages of computation. We call our approach \textit{InfoGating}. We apply InfoGating to various objectives such as multi-step forward and inverse dynamics models, Q-learning, and behavior cloning, highlighting how InfoGating can naturally help in discarding information not relevant for control. Results show that learning to identify and use minimal information can improve generalization in downstream tasks. Policies based on InfoGating are considerably more robust to irrelevant visual features, leading to improved pretraining and finetuning of RL models.
This paper presents a novel approach to multi-agent reinforcement learning (RL) for linear systems with convex polytopic constraints. Existing work on RL has demonstrated the use of model predictive control (MPC) as a function approximator for the policy and value functions. The current paper is the first work to extend this idea to the multi-agent setting. We propose the use of a distributed MPC scheme as a function approximator, with a structure allowing for distributed learning and deployment. We then show that Q-learning updates can be performed distributively without introducing nonstationarity, by reconstructing a centralized learning update. The effectiveness of the approach is demonstrated on two numerical examples.
We consider two-player zero-sum stochastic games and propose a two-timescale $Q$-learning algorithm with function approximation that is payoff-based, convergent, rational, and symmetric between the two players. In two-timescale $Q$-learning, the fast-timescale iterates are updated in spirit to the stochastic gradient descent and the slow-timescale iterates (which we use to compute the policies) are updated by taking a convex combination between its previous iterate and the latest fast-timescale iterate. Introducing the slow timescale as well as its update equation marks as our main algorithmic novelty. In the special case of linear function approximation, we establish, to the best of our knowledge, the first last-iterate finite-sample bound for payoff-based independent learning dynamics of these types. The result implies a polynomial sample complexity to find a Nash equilibrium in such stochastic games. To establish the results, we model our proposed algorithm as a two-timescale stochastic approximation and derive the finite-sample bound through a Lyapunov-based approach. The key novelty lies in constructing a valid Lyapunov function to capture the evolution of the slow-timescale iterates. Specifically, through a change of variable, we show that the update equation of the slow-timescale iterates resembles the classical smoothed best-response dynamics, where the regularized Nash gap serves as a valid Lyapunov function. This insight enables us to construct a valid Lyapunov function via a generalized variant of the Moreau envelope of the regularized Nash gap. The construction of our Lyapunov function might be of broad independent interest in studying the behavior of stochastic approximation algorithms.
Intelligent reflecting surface (IRS)-assisted unmanned aerial vehicle (UAV) communications are expected to alleviate the load of ground base stations in a cost-effective way. Existing studies mainly focus on the deployment and resource allocation of a single IRS instead of multiple IRSs, whereas it is extremely challenging for joint multi-IRS multi-user association in UAV communications with constrained reflecting resources and dynamic scenarios. To address the aforementioned challenges, we propose a new optimization algorithm for joint IRS-user association, trajectory optimization of UAVs, successive interference cancellation (SIC) decoding order scheduling and power allocation to maximize system energy efficiency. We first propose an inverse soft-Q learning-based algorithm to optimize multi-IRS multi-user association. Then, SCA and Dinkelbach-based algorithm are leveraged to optimize UAV trajectory followed by the optimization of SIC decoding order scheduling and power allocation. Finally, theoretical analysis and performance results show significant advantages of the designed algorithm in convergence rate and energy efficiency.
Parallel Reinforcement Learning (RL) frameworks are essential for mapping RL workloads to multiple computational resources, allowing for faster generation of samples, estimation of values, and policy improvement. These computational paradigms require a seamless integration of training, serving, and simulation workloads. Existing frameworks, such as Ray, are not managing this orchestration efficiently, especially in RL tasks that demand intensive input/output and synchronization between actors on a single node. In this study, we have proposed a solution implementing the reactor model, which enforces a set of actors to have a fixed communication pattern. This allows the scheduler to eliminate work needed for synchronization, such as acquiring and releasing locks for each actor or sending and processing coordination-related messages. Our framework, Lingua Franca (LF), a coordination language based on the reactor model, also supports true parallelism in Python and provides a unified interface that allows users to automatically generate dataflow graphs for RL tasks. In comparison to Ray on a single-node multi-core compute platform, LF achieves 1.21x and 11.62x higher simulation throughput in OpenAI Gym and Atari environments, reduces the average training time of synchronized parallel Q-learning by 31.2%, and accelerates multi-agent RL inference by 5.12x.
In this paper, we present a Q-learning algorithm to solve the optimal output regulation problem for discrete-time LTI systems. This off-policy algorithm only relies on using persistently exciting input-output data, measured offline. No model knowledge or state measurements are needed and the obtained optimal policy only uses past input-output information. Moreover, our formulation of the proposed algorithm renders it computationally efficient. We provide conditions that guarantee the convergence of the algorithm to the optimal solution. Finally, the performance of our method is compared to existing algorithms in the literature.
Device-to-device (D2D) communication is a promising paradigm to meet the requirement of ultra-dense, low-latency and high-rate in the fifth-generation networks. However, energy consumption is a critical issue for the D2D communication, especially for D2D relay networks. To make the best use of D2D communication, the problem of optimizing energy efficiency (EE) must be addressed. In this paper, we propose a joint power allocation and relay selection (JPARS) scheme for the improvement of energy efficiency in relay-aided D2D communications underlaying cellular network. A mixed integer nonlinear fractional programming (MINLP) problem of the total EE for D2D pairs is formulated. While ensuring the quality of service (QoS) of cellular users and D2D links, we solve the power allocation problem by Dinkelbach method and Lagrange dual decomposition. After that, Q-learning, one of the reinforcement learning algorithms, is employed to solve the relay selection problem. Finally, we provide in-depth theoretical analysis of the proposed scheme in terms of complexity and signaling overhead. Simulation results verify that the proposed scheme not only overcomes the bottleneck effect, but also nearly reaches the theoretical maximum in terms of the total EE of D2D pairs.
We study a simple model of algorithmic collusion in which Q-learning algorithms are designed in a strategic fashion. We let players (\textit{designers}) choose their exploration policy simultaneously prior to letting their algorithms repeatedly play a prisoner's dilemma. We prove that, in equilibrium, collusive behavior is reached with positive probability. Our numerical simulations indicate symmetry of the equilibria and give insight for how they are affected by a parameter of interest. We also investigate general profiles of exploration policies. We characterize the behavior of the system for extreme profiles (fully greedy and fully explorative) and use numerical simulations and clustering methods to measure the likelihood of collusive behavior in general cases.
This paper formulates a stochastic optimal control problem for linear networked control systems featuring stochastic packet disordering with a unique stabilizing solution certified. The problem is solved by proposing reinforcement learning algorithms. A measurement method is first presented to deal with PD and calculate the newest control input. The NCSs with stochastic PD are modeled as stochastic NCSs. Then, given a cost function, a modified algebraic Riccati equation is derived within the formulation. We propose offline policy iteration and value iteration algorithms to solve the MARE associated with provable convergence. These two algorithms require knowledge of NCS dynamics and PD probabilities. To release that, we further design online model-free off-policy and Q-learning algorithms with an online estimation method for PD probability. Both model-free algorithms solve the optimal control problem using real-time system states, control inputs, and PD probability estimates. Simulation results verify the proposed formulation and algorithms at last.
This study evaluates the application of a discrete action space reinforcement learning method (Q-learning) to the continuous control problem of robot inverted pendulum balancing. To speed up the learning process and to overcome technical difficulties related to the direct learning on the real robotic system, the learning phase is performed in simulation environment. A mathematical model of the system dynamics is implemented, deduced by curve fitting on data acquired from the real system. The proposed approach demonstrated feasible, featuring its application on a real world robot that learned to balance an inverted pendulum. This study also reinforces and demonstrates the importance of an accurate representation of the physical world in simulation to achieve a more efficient implementation of reinforcement learning algorithms in real world, even when using a discrete action space algorithm to control a continuous action.
In this paper, we address the problem of detecting anomalies among a given set of binary processes via learning-based controlled sensing. Each process is parameterized by a binary random variable indicating whether the process is anomalous. To identify the anomalies, the decision-making agent is allowed to observe a subset of the processes at each time instant. Also, probing each process has an associated cost. Our objective is to design a sequential selection policy that dynamically determines which processes to observe at each time with the goal to minimize the delay in making the decision and the total sensing cost. We cast this problem as a sequential hypothesis testing problem within the framework of Markov decision processes. This formulation utilizes both a Bayesian log-likelihood ratio-based reward and an entropy-based reward. The problem is then solved using two approaches: 1) a deep reinforcement learning-based approach where we design both deep Q-learning and policy gradient actor-critic algorithms; and 2) a deep active inference-based approach. Using numerical experiments, we demonstrate the efficacy of our algorithms and show that our algorithms adapt to any unknown statistical dependence pattern of the processes.
Offline multi-agent reinforcement learning is challenging due to the coupling effect of both distribution shift issue common in offline setting and the high dimension issue common in multi-agent setting, making the action out-of-distribution (OOD) and value overestimation phenomenon excessively severe. Tomitigate this problem, we propose a novel multi-agent offline RL algorithm, named CounterFactual Conservative Q-Learning (CFCQL) to conduct conservative value estimation. Rather than regarding all the agents as a high dimensional single one and directly applying single agent methods to it, CFCQL calculates conservative regularization for each agent separately in a counterfactual way and then linearly combines them to realize an overall conservative value estimation. We prove that it still enjoys the underestimation property and the performance guarantee as those single agent conservative methods do, but the induced regularization and safe policy improvement bound are independent of the agent number, which is therefore theoretically superior to the direct treatment referred to above, especially when the agent number is large. We further conduct experiments on four environments including both discrete and continuous action settings on both existing and our man-made datasets, demonstrating that CFCQL outperforms existing methods on most datasets and even with a remarkable margin on some of them.
The major challenge of offline RL is the distribution shift that appears when out-of-distribution actions are queried, which makes the policy improvement direction biased by extrapolation errors. Most existing methods address this problem by penalizing the policy or value for deviating from the behavior policy during policy improvement or evaluation. In this work, we propose a novel MISA framework to approach offline RL from the perspective of Mutual Information between States and Actions in the dataset by directly constraining the policy improvement direction. MISA constructs lower bounds of mutual information parameterized by the policy and Q-values. We show that optimizing this lower bound is equivalent to maximizing the likelihood of a one-step improved policy on the offline dataset. Hence, we constrain the policy improvement direction to lie in the data manifold. The resulting algorithm simultaneously augments the policy evaluation and improvement by adding mutual information regularizations. MISA is a general framework that unifies conservative Q-learning (CQL) and behavior regularization methods (e.g., TD3+BC) as special cases. We introduce 3 different variants of MISA, and empirically demonstrate that tighter mutual information lower bound gives better offline RL performance. In addition, our extensive experiments show MISA significantly outperforms a wide range of baselines on various tasks of the D4RL benchmark,e.g., achieving 742.9 total points on gym-locomotion tasks. Our code is available at https://github.com/sail-sg/MISA.
A compelling use case of offline reinforcement learning (RL) is to obtain a policy initialization from existing datasets followed by fast online fine-tuning with limited interaction. However, existing offline RL methods tend to behave poorly during fine-tuning. In this paper, we devise an approach for learning an effective initialization from offline data that also enables fast online fine-tuning capabilities. Our approach, calibrated Q-learning (Cal-QL), accomplishes this by learning a conservative value function initialization that underestimates the value of the learned policy from offline data, while also being calibrated, in the sense that the learned Q-values are at a reasonable scale. We refer to this property as calibration, and define it formally as providing a lower bound on the true value function of the learned policy and an upper bound on the value of some other (suboptimal) reference policy, which may simply be the behavior policy. We show that offline RL algorithms that learn such calibrated value functions lead to effective online fine-tuning, enabling us to take the benefits of offline initializations in online fine-tuning. In practice, Cal-QL can be implemented on top of the conservative Q learning (CQL) for offline RL within a one-line code change. Empirically, Cal-QL outperforms state-of-the-art methods on 9/11 fine-tuning benchmark tasks that we study in this paper. Code and video are available at https://nakamotoo.github.io/Cal-QL
Advanced vehicle control is a fundamental building block in the development of autonomous driving systems. Reinforcement learning (RL) promises to achieve control performance superior to classical approaches while keeping computational demands low during deployment. However, standard RL approaches like soft-actor critic (SAC) require extensive amounts of training data to be collected and are thus impractical for real-world application. To address this issue, we apply recently developed data-efficient deep RL methods to vehicle trajectory control. Our investigation focuses on three methods, so far unexplored for vehicle control: randomized ensemble double Q-learning (REDQ), probabilistic ensembles with trajectory sampling and model predictive path integral optimizer (PETS-MPPI), and model-based policy optimization (MBPO). We find that in the case of trajectory control, the standard model-based RL formulation used in approaches like PETS-MPPI and MBPO is not suitable. We, therefore, propose a new formulation that splits dynamics prediction and vehicle localization. Our benchmark study on the CARLA simulator reveals that the three identified data-efficient deep RL approaches learn control strategies on a par with or better than SAC, yet reduce the required number of environment interactions by more than one order of magnitude.
This paper introduces Elastic Decision Transformer (EDT), a significant advancement over the existing Decision Transformer (DT) and its variants. Although DT purports to generate an optimal trajectory, empirical evidence suggests it struggles with trajectory stitching, a process involving the generation of an optimal or near-optimal trajectory from the best parts of a set of sub-optimal trajectories. The proposed EDT differentiates itself by facilitating trajectory stitching during action inference at test time, achieved by adjusting the history length maintained in DT. Further, the EDT optimizes the trajectory by retaining a longer history when the previous trajectory is optimal and a shorter one when it is sub-optimal, enabling it to "stitch" with a more optimal trajectory. Extensive experimentation demonstrates EDT's ability to bridge the performance gap between DT-based and Q Learning-based approaches. In particular, the EDT outperforms Q Learning-based methods in a multi-task regime on the D4RL locomotion benchmark and Atari games. Videos are available at: https://kristery.github.io/edt/
Recent advances in Internet-of-Things (IoT) technologies have sparked significant interest towards developing learning-based sensing applications on embedded edge devices. These efforts, however, are being challenged by the complexities of adapting to unforeseen conditions in an open-world environment, mainly due to the intensive computational and energy demands exceeding the capabilities of edge devices. In this paper, we propose OpenSense, an open-world time-series sensing framework for making inferences from time-series sensor data and achieving incremental learning on an embedded edge device with limited resources. The proposed framework is able to achieve two essential tasks, inference and incremental learning, eliminating the necessity for powerful cloud servers. In addition, to secure enough time for incremental learning and reduce energy consumption, we need to schedule sensing activities without missing any events in the environment. Therefore, we propose two dynamic sensor scheduling techniques: (i) a class-level period assignment scheduler that finds an appropriate sensing period for each inferred class, and (ii) a Q-learning-based scheduler that dynamically determines the sensing interval for each classification moment by learning the patterns of event classes. With this framework, we discuss the design choices made to ensure satisfactory learning performance and efficient resource usage. Experimental results demonstrate the ability of the system to incrementally adapt to unforeseen conditions and to efficiently schedule to run on a resource-constrained device.