PyTorchKR Papers

최신 논문

총 1,872편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2105.01006공식 코드 · Papers with Code

Robotic Surgery With Lean Reinforcement Learning

As surgical robots become more common, automating away some of the burden of complex direct human operation becomes ever more feasible. Model-free reinforcement learning (RL) is a promising direction toward generalizable automated surgical performance, but progress has been slowed by the lack of efficient and realistic learning environments. In this paper, we describe adding reinforcement learning support to the da Vinci Skill Simulator, a training simulation used around the world to allow surgeons to learn and rehearse technical skills. We successfully teach an RL-based agent to perform sub-tasks in the simulator environment, using either image or state data. As far as we know, this is the first time an RL-based agent is taught from visual data in a surgical robotics environment. Additionally, we tackle the sample inefficiency of RL using a simple-to-implement system which we term hybrid-batch learning (HBL), effectively adding a second, long-term replay buffer to the Q-learning process. Additionally, this allows us to bootstrap learning from images from the data collected using the easier task of learning from state. We show that HBL decreases our learning times significantly.

arXiv:2105.00704공식 코드 · Papers with Code

Action Candidate Based Clipped Double Q-learning for Discrete and Continuous Action Tasks

Double Q-learning is a popular reinforcement learning algorithm in Markov decision process (MDP) problems. Clipped Double Q-learning, as an effective variant of Double Q-learning, employs the clipped double estimator to approximate the maximum expected action value. Due to the underestimation bias of the clipped double estimator, performance of clipped Double Q-learning may be degraded in some stochastic environments. In this paper, in order to reduce the underestimation bias, we propose an action candidate based clipped double estimator for Double Q-learning. Specifically, we first select a set of elite action candidates with the high action values from one set of estimators. Then, among these candidates, we choose the highest valued action from the other set of estimators. Finally, we use the maximum value in the second set of estimators to clip the action value of the chosen action in the first set of estimators and the clipped value is used for approximating the maximum expected action value. Theoretically, the underestimation bias in our clipped Double Q-learning decays monotonically as the number of the action candidates decreases. Moreover, the number of action candidates controls the trade-off between the overestimation and underestimation biases. In addition, we also extend our clipped Double Q-learning to continuous action tasks via approximating the elite continuous action candidates. We empirically verify that our algorithm can more accurately estimate the maximum expected action value on some toy environments and yield good performance on several benchmark problems.

arXiv:2105.00544

CARL-DTN: Context Adaptive Reinforcement Learning based Routing Algorithm in Delay Tolerant Network

The term Delay/Disruption-Tolerant Networks (DTN) invented to describe and cover all types of long-delay, disconnected, intermittently connected networks, where mobility and outages or scheduled contacts may be experienced. This environment is characterized by frequent network partitioning, intermittent connectivity, large or variable delay, asymmetric data rate, and low transmission reliability. There have been routing protocols developed in DTN. However, those routing algorithms are design based upon specific assumptions. The assumption makes existing algorithms suitable for specific environment scenarios. Different routing algorithm uses different relay node selection criteria to select the replication node. Too Frequently forwarding messages can result in excessive packet loss and large buffer and network overhead. On the other hand, less frequent transmission leads to a lower delivery ratio. In DTN there is a trade-off off between delivery ratio and overhead. In this study, we proposed context-adaptive reinforcement learning based routing(CARL-DTN) protocol to determine optimal replicas of the message based on the real-time density. Our routing protocol jointly uses a real-time physical context, social-tie strength, and real-time message context using fuzzy logic in the routing decision. Multi-hop forwarding probability is also considered for the relay node selection by employing Q-Learning algorithm to estimate the encounter probability between nodes and to learn about nodes available in the neighbor by discounting reward. The performance of the proposed protocol is evaluated based on various simulation scenarios. The result shows that the proposed protocol has better performance in terms of message delivery ratio and overhead.

arXiv:2104.12226

RP-DQN: An application of Q-Learning to Vehicle Routing Problems

In this paper we present a new approach to tackle complex routing problems with an improved state representation that utilizes the model complexity better than previous methods. We enable this by training from temporal differences. Specifically Q-Learning is employed. We show that our approach achieves state-of-the-art performance for autoregressive policies that sequentially insert nodes to construct solutions on the CVRP. Additionally, we are the first to tackle the MDVRP with machine learning methods and demonstrate that this problem type greatly benefits from our approach over other ML methods.

arXiv:2104.11320

Federated Double Deep Q-learning for Joint Delay and Energy Minimization in IoT networks

In this paper, we propose a federated deep reinforcement learning framework to solve a multi-objective optimization problem, where we consider minimizing the expected long-term task completion delay and energy consumption of IoT devices. This is done by optimizing offloading decisions, computation resource allocation, and transmit power allocation. Since the formulated problem is a mixed-integer non-linear programming (MINLP), we first cast our problem as a multi-agent distributed deep reinforcement learning (DRL) problem and address it using double deep Q-network (DDQN), where the actions are offloading decisions. The immediate cost of each agent is calculated through solving either the transmit power optimization or local computation resource optimization, based on the selected offloading decisions (actions). Then, to enhance the learning speed of IoT devices (agents), we incorporate federated learning (FDL) at the end of each episode. FDL enhances the scalability of the proposed DRL framework, creates a context for cooperation between agents, and minimizes their privacy concerns. Our numerical results demonstrate the efficacy of our proposed federated DDQN framework in terms of learning speed compared to federated deep Q network (DQN) and non-federated DDQN algorithms. In addition, we investigate the impact of batch size, network layers, DDQN target network update frequency on the learning speed of the FDL.

arXiv:2104.10403

Model-aided Deep Reinforcement Learning for Sample-efficient UAV Trajectory Design in IoT Networks

Deep Reinforcement Learning (DRL) is gaining attention as a potential approach to design trajectories for autonomous unmanned aerial vehicles (UAV) used as flying access points in the context of cellular or Internet of Things (IoT) connectivity. DRL solutions offer the advantage of on-the-go learning hence relying on very little prior contextual information. A corresponding drawback however lies in the need for many learning episodes which severely restricts the applicability of such approach in real-world time- and energy-constrained missions. Here, we propose a model-aided deep Q-learning approach that, in contrast to previous work, considerably reduces the need for extensive training data samples, while still achieving the overarching goal of DRL, i.e to guide a battery-limited UAV on an efficient data harvesting trajectory, without prior knowledge of wireless channel characteristics and limited knowledge of wireless node locations. The key idea consists in using a small subset of nodes as anchors (i.e. with known location) and learning a model of the propagation environment while implicitly estimating the positions of regular nodes. Interaction with the model allows us to train a deep Q-network (DQN) to approximate the optimal UAV control policy. We show that in comparison with standard DRL approaches, the proposed model-aided approach requires at least one order of magnitude less training data samples to reach identical data collection performance, hence offering a first step towards making DRL a viable solution to the problem.

arXiv:2104.10455

Reinforcement Learning for Traffic Signal Control: Comparison with Commercial Systems

Recently, Intelligent Transportation Systems are leveraging the power of increased sensory coverage and computing power to deliver data-intensive solutions achieving higher levels of performance than traditional systems. Within Traffic Signal Control (TSC), this has allowed the emergence of Machine Learning (ML) based systems. Among this group, Reinforcement Learning (RL) approaches have performed particularly well. Given the lack of industry standards in ML for TSC, literature exploring RL often lacks comparison against commercially available systems and straightforward formulations of how the agents operate. Here we attempt to bridge that gap. We propose three different architectures for TSC RL agents and compare them against the currently used commercial systems MOVA, SurTrac and Cyclic controllers and provide pseudo-code for them. The agents use variations of Deep Q-Learning and Actor Critic, using states and rewards based on queue lengths. Their performance is compared in across different map scenarios with variable demand, assessing them in terms of the global delay and average queue length. We find that the RL-based systems can significantly and consistently achieve lower delays when compared with existing commercial systems.

arXiv:2104.08940

A Simulated Experiment to Explore Robotic Dialogue Strategies for People with Dementia

People with Alzheimer's disease and related dementias (ADRD) often show the problem of repetitive questioning, which brings a great burden on persons with ADRD (PwDs) and their caregivers. Conversational robots hold promise of coping with this problem and hence alleviating the burdens on caregivers. In this paper, we proposed a partially observable markov decision process (POMDP) model for the PwD-robot interaction in the context of repetitive questioning, and used Q-learning to learn an adaptive conversation strategy (i.e., rate of follow-up question and difficulty of follow-up question) towards PwDs with different cognitive capabilities and different engagement levels. The results indicated that Q-learning was helpful for action selection for the robot. This may be a useful step towards the application of conversational social robots to cope with repetitive questioning in PwDs.

arXiv:2104.08805공식 코드 · Papers with Code

Low-rank State-action Value-function Approximation

Value functions are central to Dynamic Programming and Reinforcement Learning but their exact estimation suffers from the curse of dimensionality, challenging the development of practical value-function (VF) estimation algorithms. Several approaches have been proposed to overcome this issue, from non-parametric schemes that aggregate states or actions to parametric approximations of state and action VFs via, e.g., linear estimators or deep neural networks. Relevantly, several high-dimensional state problems can be well-approximated by an intrinsic low-rank structure. Motivated by this and leveraging results from low-rank optimization, this paper proposes different stochastic algorithms to estimate a low-rank factorization of the $Q(s, a)$ matrix. This is a non-parametric alternative to VF approximation that dramatically reduces the computational and sample complexities relative to classical $Q$-learning methods that estimate $Q(s,a)$ separately for each state-action pair.

arXiv:2104.07749

Actionable Models: Unsupervised Offline Reinforcement Learning of Robotic Skills

We consider the problem of learning useful robotic skills from previously collected offline data without access to manually specified rewards or additional online exploration, a setting that is becoming increasingly important for scaling robot learning by reusing past robotic data. In particular, we propose the objective of learning a functional understanding of the environment by learning to reach any goal state in a given dataset. We employ goal-conditioned Q-learning with hindsight relabeling and develop several techniques that enable training in a particularly challenging offline setting. We find that our method can operate on high-dimensional camera images and learn a variety of skills on real robots that generalize to previously unseen scenes and objects. We also show that our method can learn to reach long-horizon goals across multiple episodes through goal chaining, and learn rich representations that can help with downstream tasks through pre-training or auxiliary objectives. The videos of our experiments can be found at https://actionable-models.github.io

arXiv:2008.03336

Evaluating Load Models and Their Impacts on Power Transfer Limits

Power transfer limits or transfer capability (TC) directly relate to the system operation and control as well as electricity markets. As a consequence, their assessment has to comply with static constraints, such as line thermal limits, and dynamic constraints, such as transient stability limits, voltage stability limits and small-signal stability limits. Since the load dynamics have substantial impacts on power system transient stability, load models are one critical factor that affects the power transfer limits. Currently, multiple load models have been proposed and adopted in the industry and academia, including the ZIP model, ZIP plus induction motor composite model (ZIP + IM) and WECC composite load model (WECC CLM). Each of them has its unique advantages, but their impacts on the power transfer limits are not yet adequately addressed. One existing challenge is fitting the high-order nonlinear models such as WECC CLM. In this study, we innovatively adopt double deep Q-learning Network (DDQN) agent as a general load modeling tool in the dynamic assessment procedure and fit the same transient field measurements into different load models. A comprehensive evaluation is then conducted to quantify the load models' impacts on the power transfer limits. The simulation environment is the IEEE-39 bus system constructed in Transient Security Assessment Tool (TSAT).

arXiv:2010.02749

Machine Learning Empowered Trajectory and Passive Beamforming Design in UAV-RIS Wireless Networks

A novel framework is proposed for integrating reconfigurable intelligent surfaces (RIS) in unmanned aerial vehicle (UAV) enabled wireless networks, where an RIS is deployed for enhancing the service quality of the UAV. Non-orthogonal multiple access (NOMA) technique is invoked to further improve the spectrum efficiency of the network, while mobile users (MUs) are considered as roaming continuously. The energy consumption minimizing problem is formulated by jointly designing the movement of the UAV, phase shifts of the RIS, power allocation policy from the UAV to MUs, as well as determining the dynamic decoding order. A decaying deep Q-network (D-DQN) based algorithm is proposed for tackling this pertinent problem. In the proposed D-DQN based algorithm, the central controller is selected as an agent for periodically observing the state of UAV-enabled wireless network and for carrying out actions to adapt to the dynamic environment. In contrast to the conventional DQN algorithm, the decaying learning rate is leveraged in the proposed D-DQN based algorithm for attaining a tradeoff between accelerating training speed and converging to the local optimal. Numerical results demonstrate that: 1) In contrast to the conventional Q-learning algorithm, which cannot converge when being adopted for solving the formulated problem, the proposed D-DQN based algorithm is capable of converging with minor constraints; 2) The energy dissipation of the UAV can be significantly reduced by integrating RISs in UAV-enabled wireless networks; 3) By designing the dynamic decoding order and power allocation policy, the RIS-NOMA case consumes 11.7% less energy than the RIS-OMA case.

arXiv:2101.00270

Deep Reinforcement Learning-based Anti-jamming Power Allocation in a Two-cell NOMA Network

The performance of Non-orthogonal Multiple Access (NOMA) system dramatically decreases in the presence of inter-cell interference. This condition gets more challenging if a smart jammer is interacting in a network. In this paper, the NOMA power allocation of two independent Base Stations (BSs) against a smart jammer is, modeled as a sequential game. In this game, at first, each BS as a leader independently chooses its power allocation strategy. Then, the smart jammer as the follower selects its optimal strategy based on the strategies of the BSs. The solutions of this game are, derived under different conditions. Based on the game-theoretical analysis, three new schemes are proposed for anti-jamming NOMA power allocation in a two-cell scenario called a) Q-Learning based Unselfish (QLU) NOMA power allocation scheme, b) Deep Q-Learning based Unselfish (DQLU) NOMA power allocation scheme, and c) Hot Booting Deep Q-Learning based Unselfish (HBDQLU) NOMA power allocation scheme. In these methods the BSs do not coordinate with each other. But our analysis theoretically predicts that with high probability, the proposed methods will converge to the optimal strategy from the total network point of view. Simulation results show the convergence of the proposed schemes and also their outperformance with respect to the Q-Learning-based Selfish (QLS) NOMA power allocation method.

arXiv:2104.05311

Prospect-theoretic Q-learning

We consider a prospect theoretic version of the classical Q-learning algorithm for discounted reward Markov decision processes, wherein the controller perceives a distorted and noisy future reward, modeled by a nonlinearity that accentuates gains and underrepresents losses relative to a reference point. We analyze the asymptotic behavior of the scheme by analyzing its limiting differential equation and using the theory of monotone dynamical systems to infer its asymptotic behavior. Specifically, we show convergence to equilibria, and establish some qualitative facts about the equilibria themselves.

arXiv:2104.04848공식 코드 · Papers with Code

Autoequivariant Network Search via Group Decomposition

Recent works show that group equivariance as an inductive bias improves neural network performance for both classification and generation. However, designing group-equivariant neural networks is challenging when the group of interest is large and is unknown. Moreover, inducing equivariance can significantly reduce the number of independent parameters in a network with fixed feature size, affecting its overall performance. We address these problems by proving a new group-theoretic result in the context of equivariant neural networks that shows that a network is equivariant to a large group if and only if it is equivariant to smaller groups from which it is constructed. Using this result, we design a novel fast group equivariant construction algorithm, and a deep Q-learning-based search algorithm in a reduced search space, yielding what we call autoequivariant networks (AENs). AENs find the right balance between equivariance and network size when tested on new benchmark datasets, G-MNIST and G-Fashion-MNIST, obtained via group transformations on MNIST and Fashion-MNIST respectively that we release. Extending these results to group convolutional neural networks, where we optimize between equivariances, augmentations, and network sizes, we find group equivariance to be the most dominating factor in all high-performing GCNNs on several datasets like CIFAR10, SVHN, RotMNIST, ASL, EMNIST, and KMNIST.

arXiv:2104.04036공식 코드 · Papers with Code

Optimal Market Making by Reinforcement Learning

We apply Reinforcement Learning algorithms to solve the classic quantitative finance Market Making problem, in which an agent provides liquidity to the market by placing buy and sell orders while maximizing a utility function. The optimal agent has to find a delicate balance between the price risk of her inventory and the profits obtained by capturing the bid-ask spread. We design an environment with a reward function that determines an order relation between policies equivalent to the original utility function. When comparing our agents with the optimal solution and a benchmark symmetric agent, we find that the Deep Q-Learning algorithm manages to recover the optimal agent.

arXiv:2104.03153

Towards Resilience for Multi-Agent $QD$-Learning

This paper considers the multi-agent reinforcement learning (MARL) problem for a networked (peer-to-peer) system in the presence of Byzantine agents. We build on an existing distributed $Q$-learning algorithm, and allow certain agents in the network to behave in an arbitrary and adversarial manner (as captured by the Byzantine attack model). Under the proposed algorithm, if the network topology is $(2F+1)$-robust and up to $F$ Byzantine agents exist in the neighborhood of each regular agent, we establish the almost sure convergence of all regular agents' value functions to the neighborhood of the optimal value function of all regular agents. For each state, if the optimal $Q$-values of all regular agents corresponding to different actions are sufficiently separated, our approach allows each regular agent to learn the optimal policy for all regular agents.

arXiv:2104.02059

Distributed Deep Reinforcement Learning for Collaborative Spectrum Sharing

Spectrum sharing among users is a fundamental problem in the management of any wireless network. In this paper, we discuss the problem of distributed spectrum collaboration without central management under general unknown channels. Since the cost of communication, coordination and control is rapidly increasing with the number of devices and the expanding bandwidth used there is an obvious need to develop distributed techniques for spectrum collaboration where no explicit signaling is used. In this paper, we combine game-theoretic insights with deep Q-learning to provide a novel asymptotically optimal solution to the spectrum collaboration problem. We propose a deterministic distributed deep reinforcement learning(D3RL) mechanism using a deep Q-network (DQN). It chooses the channels using the Q-values and the channel loads while limiting the options available to the user to a few channels with the highest Q-values and among those, it selects the least loaded channel. Using insights from both game theory and combinatorial optimization we show that this technique is asymptotically optimal for large overloaded networks. The selected channel and the outcome of the successful transmission are fed back into the learning of the deep Q-network to incorporate it into the learning of the Q-values. We also analyzed performance to understand the behavior of D3RL in differ

arXiv:2104.01646

SOLO: Search Online, Learn Offline for Combinatorial Optimization Problems

We study combinatorial problems with real world applications such as machine scheduling, routing, and assignment. We propose a method that combines Reinforcement Learning (RL) and planning. This method can equally be applied to both the offline, as well as online, variants of the combinatorial problem, in which the problem components (e.g., jobs in scheduling problems) are not known in advance, but rather arrive during the decision-making process. Our solution is quite generic, scalable, and leverages distributional knowledge of the problem parameters. We frame the solution process as an MDP, and take a Deep Q-Learning approach wherein states are represented as graphs, thereby allowing our trained policies to deal with arbitrary changes in a principled manner. Though learned policies work well in expectation, small deviations can have substantial negative effects in combinatorial settings. We mitigate these drawbacks by employing our graph-convolutional policies as non-optimal heuristics in a compatible search algorithm, Monte Carlo Tree Search, to significantly improve overall performance. We demonstrate our method on two problems: Machine Scheduling and Capacitated Vehicle Routing. We show that our method outperforms custom-tailored mathematical solvers, state of the art learning-based algorithms, and common heuristics, both in computation time and performance.

arXiv:2103.04154

Decentralized Microgrid Energy Management: A Multi-agent Correlated Q-learning Approach

Microgrids (MG) are anticipated to be important players in the future smart grid. For proper operation of MGs an Energy Management System (EMS) is essential. The EMS of an MG could be rather complicated when renewable energy resources (RER), energy storage system (ESS) and demand side management (DSM) need to be orchestrated. Furthermore, these systems may belong to different entities and competition may exist between them. Nash equilibrium is most commonly used for coordination of such entities however the convergence and existence of Nash equilibrium can not always be guaranteed. To this end, we use the correlated equilibrium to coordinate agents, whose convergence can be guaranteed. In this paper, we build an energy trading model based on mid-market rate, and propose a correlated Q-learning (CEQ) algorithm to maximize the revenue of each agent. Our results show that CEQ is able to balance the revenue of agents without harming total benefit. In addition, compared with Q-learning without correlation, CEQ could save 19.3% cost for the DSM agent and 44.2% more benefits for the ESS agent.