PyTorchKR Papers

최신 논문

총 1,872편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2202.02850

Stochastic Gradient Descent with Dependent Data for Offline Reinforcement Learning

In reinforcement learning (RL), offline learning decoupled learning from data collection and is useful in dealing with exploration-exploitation tradeoff and enables data reuse in many applications. In this work, we study two offline learning tasks: policy evaluation and policy learning. For policy evaluation, we formulate it as a stochastic optimization problem and show that it can be solved using approximate stochastic gradient descent (aSGD) with time-dependent data. We show aSGD achieves $\tilde O(1/t)$ convergence when the loss function is strongly convex and the rate is independent of the discount factor $\gamma$. This result can be extended to include algorithms making approximately contractive iterations such as TD(0). The policy evaluation algorithm is then combined with the policy iteration algorithm to learn the optimal policy. To achieve an $\epsilon$ accuracy, the complexity of the algorithm is $\tilde O(\epsilon^{-2}(1-\gamma)^{-5})$, which matches the complexity bound for classic online RL algorithms such as Q-learning.

arXiv:2201.12385

A deep Q-learning method for optimizing visual search strategies in backgrounds of dynamic noise

Humans process visual information with varying resolution (foveated visual system) and explore images by orienting through eye movements the high-resolution fovea to points of interest. The Bayesian ideal searcher (IS) that employs complete knowledge of task-relevant information optimizes eye movement strategy and achieves the optimal search performance. The IS can be employed as an important tool to evaluate the optimality of human eye movements, and potentially provide guidance to improve human observer visual search strategies. Najemnik and Geisler (2005) derived an IS for backgrounds of spatial 1/f noise. The corresponding template responses follow Gaussian distributions and the optimal search strategy can be analytically determined. However, the computation of the IS can be intractable when considering more realistic and complex backgrounds such as medical images. Modern reinforcement learning methods, successfully applied to obtain optimal policy for a variety of tasks, do not require complete knowledge of the background generating functions and can be potentially applied to anatomical backgrounds. An important first step is to validate the optimality of the reinforcement learning method. In this study, we investigate the ability of a reinforcement learning method that employs Q-network to approximate the IS. We demonstrate that the search strategy corresponding to the Q-network is consistent with the IS search strategy. The findings show the potential of the reinforcement learning with Q-network approach to estimate optimal eye movement planning with real anatomical backgrounds.

arXiv:2201.09754

Deep Reinforcement Learning with Spiking Q-learning

With the help of special neuromorphic hardware, spiking neural networks (SNNs) are expected to realize artificial intelligence (AI) with less energy consumption. It provides a promising energy-efficient way for realistic control tasks by combining SNNs with deep reinforcement learning (RL). There are only a few existing SNN-based RL methods at present. Most of them either lack generalization ability or employ Artificial Neural Networks (ANNs) to estimate value function in training. The former needs to tune numerous hyper-parameters for each scenario, and the latter limits the application of different types of RL algorithm and ignores the large energy consumption in training. To develop a robust spike-based RL method, we draw inspiration from non-spiking interneurons found in insects and propose the deep spiking Q-network (DSQN), using the membrane voltage of non-spiking neurons as the representation of Q-value, which can directly learn robust policies from high-dimensional sensory inputs using end-to-end RL. Experiments conducted on 17 Atari games demonstrate the DSQN is effective and even outperforms the ANN-based deep Q-network (DQN) in most games. Moreover, the experiments show superior learning stability and robustness to adversarial attacks of DSQN.

arXiv:2201.08610공식 코드 · Papers with Code

Deep Q-learning: a robust control approach

In this paper, we place deep Q-learning into a control-oriented perspective and study its learning dynamics with well-established techniques from robust control. We formulate an uncertain linear time-invariant model by means of the neural tangent kernel to describe learning. We show the instability of learning and analyze the agent's behavior in frequency-domain. Then, we ensure convergence via robust controllers acting as dynamical rewards in the loss function. We synthesize three controllers: state-feedback gain scheduling H2, dynamic Hinf, and constant gain Hinf controllers. Setting up the learning agent with a control-oriented tuning methodology is more transparent and has well-established literature compared to the heuristics in reinforcement learning. In addition, our approach does not use a target network and randomized replay memory. The role of the target network is overtaken by the control input, which also exploits the temporal dependency of samples (opposed to a randomized memory buffer). Numerical simulations in different OpenAI Gym environments suggest that the Hinf controlled learning performs slightly better than Double deep Q-learning.

arXiv:2201.08518

Optimal variance-reduced stochastic approximation in Banach spaces

We study the problem of estimating the fixed point of a contractive operator defined on a separable Banach space. Focusing on a stochastic query model that provides noisy evaluations of the operator, we analyze a variance-reduced stochastic approximation scheme, and establish non-asymptotic bounds for both the operator defect and the estimation error, measured in an arbitrary semi-norm. In contrast to worst-case guarantees, our bounds are instance-dependent, and achieve the local asymptotic minimax risk non-asymptotically. For linear operators, contractivity can be relaxed to multi-step contractivity, so that the theory can be applied to problems like average reward policy evaluation problem in reinforcement learning. We illustrate the theory via applications to stochastic shortest path problems, two-player zero-sum Markov games, as well as policy evaluation and $Q$-learning for tabular Markov decision processes.

arXiv:2201.08078공식 코드 · Papers with Code

Addressing Maximization Bias in Reinforcement Learning with Two-Sample Testing

Value-based reinforcement-learning algorithms have shown strong results in games, robotics, and other real-world applications. Overestimation bias is a known threat to those algorithms and can sometimes lead to dramatic performance decreases or even complete algorithmic failure. We frame the bias problem statistically and consider it an instance of estimating the maximum expected value (MEV) of a set of random variables. We propose the $T$-Estimator (TE) based on two-sample testing for the mean, that flexibly interpolates between over- and underestimation by adjusting the significance level of the underlying hypothesis tests. We also introduce a generalization, termed $K$-Estimator (KE), that obeys the same bias and variance bounds as the TE and relies on a nearly arbitrary kernel function. We introduce modifications of $Q$-Learning and the Bootstrapped Deep $Q$-Network (BDQN) using the TE and the KE, and prove convergence in the tabular setting. Furthermore, we propose an adaptive variant of the TE-based BDQN that dynamically adjusts the significance level to minimize the absolute estimation bias. All proposed estimators and algorithms are thoroughly tested and validated on diverse tasks and environments, illustrating the bias control and performance potential of the TE and KE.

arXiv:2201.05034

Criticality-Based Varying Step-Number Algorithm for Reinforcement Learning

In the context of reinforcement learning we introduce the concept of criticality of a state, which indicates the extent to which the choice of action in that particular state influences the expected return. That is, a state in which the choice of action is more likely to influence the final outcome is considered as more critical than a state in which it is less likely to influence the final outcome. We formulate a criticality-based varying step number algorithm (CVS) - a flexible step number algorithm that utilizes the criticality function provided by a human, or learned directly from the environment. We test it in three different domains including the Atari Pong environment, Road-Tree environment, and Shooter environment. We demonstrate that CVS is able to outperform popular learning algorithms such as Deep Q-Learning and Monte Carlo.

arXiv:2201.02787

Age-of-information minimization via opportunistic sampling by an energy harvesting source

Herein, minimization of time-averaged age-of-information (AoI) in an energy harvesting (EH) source setting is considered. The EH source opportunistically samples one or multiple processes over discrete time instants and sends the status updates to a sink node over a wireless fading channel. Each time, the EH node decides whether to probe the link quality and then decides whether to sample a process and communicate based on the channel probe outcome. The trade-off is between the freshness of information available at the sink node and the available energy at the source node. We use infinite horizon Markov decision process (MDP) to formulate the AoI minimization problem for two scenarios where energy arrival and channel fading processes are: (i) independent and identically distributed (i.i.d.), (ii) Markovian. In i.i.d. setting, after channel probing, the optimal source sampling policy is shown to be a threshold policy. Also, for unknown channel state and EH characteristics, a variant of the Q-learning algorithm is proposed for the two-stage action model, that seeks to learn the optimal policy. For Markovian system, the problem is again formulated as an MDP, and a learning algorithm is provided for unknown dynamics. Finally, numerical results demonstrate the policy structures and performance trade-offs.

arXiv:2201.02058

Sales Time Series Analytics Using Deep Q-Learning

The article describes the use of deep Q-learning models in the problems of sales time series analytics. In contrast to supervised machine learning which is a kind of passive learning using historical data, Q-learning is a kind of active learning with goal to maximize a reward by optimal sequence of actions. Model free Q-learning approach for optimal pricing strategies and supply-demand problems was considered in the work. The main idea of the study is to show that using deep Q-learning approach in time series analytics, the sequence of actions can be optimized by maximizing the reward function when the environment for learning agent interaction can be modeled using the parametric model and in the case of using the model which is based on the historical data. In the pricing optimizing case study environment was modeled using sales dependence on extras price and randomly simulated demand. In the pricing optimizing case study, the environment was modeled using sales dependence on extra price and randomly simulated demand. In the supply-demand case study, it was proposed to use historical demand time series for environment modeling, agent states were represented by promo actions, previous demand values and weekly seasonality features. Obtained results show that using deep Q-learning, we can optimize the decision making process for price optimization and supply-demand problems. Environment modeling using parametric models and historical data can be used for the cold start of learning agent. On the next steps, after the cold start, the trained agent can be used in real business environment.

arXiv:2112.08670

Amortized Noisy Channel Neural Machine Translation

Noisy channel models have been especially effective in neural machine translation (NMT). However, recent approaches like "beam search and rerank" (BSR) incur significant computation overhead during inference, making real-world application infeasible. We aim to study if it is possible to build an amortized noisy channel NMT model such that when we do greedy decoding during inference, the translation accuracy matches that of BSR in terms of reward (based on the source-to-target log probability and the target-to-source log probability) and quality (based on BLEU and BLEURT). We attempt three approaches to train the new model: knowledge distillation, one-step-deviation imitation learning, and Q learning. The first approach obtains the noisy channel signal from a pseudo-corpus, and the latter two approaches aim to optimize toward a noisy-channel MT reward directly. For all three approaches, the generated translations fail to achieve rewards comparable to BSR, but the translation quality approximated by BLEU and BLEURT is similar to the quality of BSR-produced translations. Additionally, all three approaches speed up inference by 1-2 orders of magnitude.

arXiv:2201.00236

Operator Deep Q-Learning: Zero-Shot Reward Transferring in Reinforcement Learning

Reinforcement learning (RL) has drawn increasing interests in recent years due to its tremendous success in various applications. However, standard RL algorithms can only be applied for single reward function, and cannot adapt to an unseen reward function quickly. In this paper, we advocate a general operator view of reinforcement learning, which enables us to directly approximate the operator that maps from reward function to value function. The benefit of learning the operator is that we can incorporate any new reward function as input and attain its corresponding value function in a zero-shot manner. To approximate this special type of operator, we design a number of novel operator neural network architectures based on its theoretical properties. Our design of operator networks outperform the existing methods and the standard design of general purpose operator network, and we demonstrate the benefit of our operator deep Q-learning framework in several tasks including reward transferring for offline policy evaluation (OPE) and reward transferring for offline policy optimization in a range of tasks.

arXiv:2201.00286

Reinforcement Learning for Task Specifications with Action-Constraints

In this paper, we use concepts from supervisory control theory of discrete event systems to propose a method to learn optimal control policies for a finite-state Markov Decision Process (MDP) in which (only) certain sequences of actions are deemed unsafe (respectively safe). We assume that the set of action sequences that are deemed unsafe and/or safe are given in terms of a finite-state automaton; and propose a supervisor that disables a subset of actions at every state of the MDP so that the constraints on action sequence are satisfied. Then we present a version of the Q-learning algorithm for learning optimal policies in the presence of non-Markovian action-sequence and state constraints, where we use the development of reward machines to handle the state constraints. We illustrate the method using an example that captures the utility of automata-based methods for non-Markovian state and action specifications for reinforcement learning and show the results of simulations in this setting.

arXiv:2112.14843

A Graph Attention Learning Approach to Antenna Tilt Optimization

6G will move mobile networks towards increasing levels of complexity. To deal with this complexity, optimization of network parameters is key to ensure high performance and timely adaptivity to dynamic network environments. The optimization of the antenna tilt provides a practical and cost-efficient method to improve coverage and capacity in the network. Previous methods based on Reinforcement Learning (RL) have shown great promise for tilt optimization by learning adaptive policies outperforming traditional tilt optimization methods. However, most existing RL methods are based on single-cell features representation, which fails to fully characterize the agent state, resulting in suboptimal performance. Also, most of such methods lack scalability, due to state-action explosion, and generalization ability. In this paper, we propose a Graph Attention Q-learning (GAQ) algorithm for tilt optimization. GAQ relies on a graph attention mechanism to select relevant neighbors information, improve the agent state representation, and update the tilt control policy based on a history of observations using a Deep Q-Network (DQN). We show that GAQ efficiently captures important network information and outperforms standard DQN with local information by a large margin. In addition, we demonstrate its ability to generalize to network deployments of different sizes and densities.

arXiv:2112.14582공식 코드 · Papers with Code

A Statistical Analysis of Polyak-Ruppert Averaged Q-learning

We study Q-learning with Polyak-Ruppert averaging in a discounted Markov decision process in synchronous and tabular settings. Under a Lipschitz condition, we establish a functional central limit theorem for the averaged iteration $\bar{\boldsymbol{Q}}_T$ and show that its standardized partial-sum process converges weakly to a rescaled Brownian motion. The functional central limit theorem implies a fully online inference method for reinforcement learning. Furthermore, we show that $\bar{\boldsymbol{Q}}_T$ is the regular asymptotically linear (RAL) estimator for the optimal Q-value function $\boldsymbol{Q}^*$ that has the most efficient influence function. We present a nonasymptotic analysis for the $\ell_{\infty}$ error, $\mathbb{E}\|\bar{\boldsymbol{Q}}_T-\boldsymbol{Q}^*\|_{\infty}$, showing that it matches the instance-dependent lower bound for polynomial step sizes. Similar results are provided for entropy-regularized Q-learning without the Lipschitz condition.

arXiv:2112.13267공식 코드 · Papers with Code

Task and Model Agnostic Adversarial Attack on Graph Neural Networks

Adversarial attacks on Graph Neural Networks (GNNs) reveal their security vulnerabilities, limiting their adoption in safety-critical applications. However, existing attack strategies rely on the knowledge of either the GNN model being used or the predictive task being attacked. Is this knowledge necessary? For example, a graph may be used for multiple downstream tasks unknown to a practical attacker. It is thus important to test the vulnerability of GNNs to adversarial perturbations in a model and task agnostic setting. In this work, we study this problem and show that GNNs remain vulnerable even when the downstream task and model are unknown. The proposed algorithm, TANDIS (Targeted Attack via Neighborhood DIStortion) shows that distortion of node neighborhoods is effective in drastically compromising prediction performance. Although neighborhood distortion is an NP-hard problem, TANDIS designs an effective heuristic through a novel combination of Graph Isomorphism Network with deep Q-learning. Extensive experiments on real datasets and state-of-the-art models show that, on average, TANDIS is up to 50% more effective than state-of-the-art techniques, while being more than 1000 times faster.

arXiv:2112.12288공식 코드 · Papers with Code

Safety and Liveness Guarantees through Reach-Avoid Reinforcement Learning

Reach-avoid optimal control problems, in which the system must reach certain goal conditions while staying clear of unacceptable failure modes, are central to safety and liveness assurance for autonomous robotic systems, but their exact solutions are intractable for complex dynamics and environments. Recent successes in reinforcement learning methods to approximately solve optimal control problems with performance objectives make their application to certification problems attractive; however, the Lagrange-type objective used in reinforcement learning is not suitable to encode temporal logic requirements. Recent work has shown promise in extending the reinforcement learning machinery to safety-type problems, whose objective is not a sum, but a minimum (or maximum) over time. In this work, we generalize the reinforcement learning formulation to handle all optimal control problems in the reach-avoid category. We derive a time-discounted reach-avoid Bellman backup with contraction mapping properties and prove that the resulting reach-avoid Q-learning algorithm converges under analogous conditions to the traditional Lagrange-type problem, yielding an arbitrarily tight conservative approximation to the reach-avoid set. We further demonstrate the use of this formulation with deep reinforcement learning methods, retaining zero-violation guarantees by treating the approximate solutions as untrusted oracles in a model-predictive supervisory control framework. We evaluate our proposed framework on a range of nonlinear systems, validating the results against analytic and numerical solutions, and through Monte Carlo simulation in previously intractable problems. Our results open the door to a range of learning-based methods for safe-and-live autonomous behavior, with applications across robotics and automation. See https://github.com/SafeRoboticsLab/safety_rl for code and supplementary material.

arXiv:2112.11090

Aerial Base Station Positioning and Power Control for Securing Communications: A Deep Q-Network Approach

The unmanned aerial vehicle (UAV) is one of the technological breakthroughs that supports a variety of services, including communications. UAV will play a critical role in enhancing the physical layer security of wireless networks. This paper defines the problem of eavesdropping on the link between the ground user and the UAV, which serves as an aerial base station (ABS). The reinforcement learning algorithms Q-learning and deep Q-network (DQN) are proposed for optimizing the position of the ABS and the transmission power to enhance the data rate of the ground user. This increases the secrecy capacity without the system knowing the location of the eavesdropper. Simulation results show fast convergence and the highest secrecy capacity of the proposed DQN compared to Q-learning and baseline approaches.

arXiv:2112.07859

Finite-Sample Analysis of Decentralized Q-Learning for Stochastic Games

Learning in stochastic games is arguably the most standard and fundamental setting in multi-agent reinforcement learning (MARL). In this paper, we consider decentralized MARL in stochastic games in the non-asymptotic regime. In particular, we establish the finite-sample complexity of fully decentralized Q-learning algorithms in a significant class of general-sum stochastic games (SGs) - weakly acyclic SGs, which includes the common cooperative MARL setting with an identical reward to all agents (a Markov team problem) as a special case. We focus on the practical while challenging setting of fully decentralized MARL, where neither the rewards nor the actions of other agents can be observed by each agent. In fact, each agent is completely oblivious to the presence of other decision makers. Both the tabular and the linear function approximation cases have been considered. In the tabular setting, we analyze the sample complexity for the decentralized Q-learning algorithm to converge to a Markov perfect equilibrium (Nash equilibrium). With linear function approximation, the results are for convergence to a linear approximated equilibrium - a new notion of equilibrium that we propose - which describes that each agent's policy is a best reply (to other agents) within a linear space. Numerical experiments are also provided for both settings to demonstrate the results.

arXiv:2112.07031

Teaching a Robot to Walk Using Reinforcement Learning

Classical control techniques such as PID and LQR have been used effectively in maintaining a system state, but these techniques become more difficult to implement when the model dynamics increase in complexity and sensitivity. For adaptive robotic locomotion tasks with several degrees of freedom, this task becomes infeasible with classical control techniques. Instead, reinforcement learning can train optimal walking policies with ease. We apply deep Q-learning and augmented random search (ARS) to teach a simulated two-dimensional bipedal robot how to walk using the OpenAI Gym BipedalWalker-v3 environment. Deep Q-learning did not yield a high reward policy, often prematurely converging to suboptimal local maxima likely due to the coarsely discretized action space. ARS, however, resulted in a better trained robot, and produced an optimal policy which officially "solves" the BipedalWalker-v3 problem. Various naive policies, including a random policy, a manually encoded inch forward policy, and a stay still policy, were used as benchmarks to evaluate the proficiency of the learning algorithm results.

arXiv:2112.06018

Control-Tutored Reinforcement Learning: Towards the Integration of Data-Driven and Model-Based Control

We present an architecture where a feedback controller derived on an approximate model of the environment assists the learning process to enhance its data efficiency. This architecture, which we term as Control-Tutored Q-learning (CTQL), is presented in two alternative flavours. The former is based on defining the reward function so that a Boolean condition can be used to determine when the control tutor policy is adopted, while the latter, termed as probabilistic CTQL (pCTQL), is instead based on executing calls to the tutor with a certain probability during learning. Both approaches are validated, and thoroughly benchmarked against Q-Learning, by considering the stabilization of an inverted pendulum as defined in OpenAI Gym as a representative problem.