This paper surveys the field of reinforcement learning from a computer-science perspective. It is written to be accessible to researchers familiar with machine learning. Both the historical basis of the field and a broad selection of current work are summarized. Reinforcement learning is the problem faced by an agent that learns behavior through trial-and-error interactions with a dynamic environment. The work described here has a resemblance to work in psychology, but differs considerably in the details and in the use of the word "reinforcement." The paper discusses central issues of reinforcement learning, including trading off exploration and exploitation, establishing the foundations of the field via Markov decision theory, learning from delayed reinforcement, constructing empirical models to accelerate learning, making use of generalization and hierarchy, and coping with hidden state. It concludes with a survey of some implemented systems and an assessment of the practical utility of current methods for reinforcement learning.

Abstract not found

We present a new algorithm, Prioritized Sweeping, for e cient prediction and control of stochas-tic Markov systems. Incremental learning methods such asTemporal Di erencing and Q-learning have fast real time performance. Classical methods are slower, but more accurate, because they make full use of the observations. Prioritized Sweeping aims for the best of both worlds. It uses all previous experiences both to prioritize important dynamic programming sweeps and to guide the exploration of state-space. We compare Prioritized Sweeping with other reinforcement learning schemes for a number of di erent stochastic optimal control prob-lems. It successfully solves large state-space real time problems with which other methods have di culty. 1 1

Abstract. To date, reinforcement learning has mostly been studied solving simple learning tasks. Reinforcement learning methods that have been studied so far typically converge slowly. The purpose of this work is thus two-fold: 1) to investigate the utility of reinforcement learning in solving much more complicated learning tasks than previously studied, and 2) to investigate methods that will speed up reinforcement learning. This paper compares eight reinforcement learning frameworks: adaptive heuristic critic (AHC) learning due to Sutton, Q-learning due to Watkins, and three extensions to both basic methods for speeding up learning. The three extensions are experience replay, learning action models for planning, and teaching. The frameworks were investigated using connectionism as an approach to generalization. To evaluate the performance of different frame-works, a dynamic environment was used as a testbed. The enviromaaent is moderately complex and nondetermin-istic. This paper describes these frameworks and algorithms in detail and presents empirical evaluation of the frameworks.

Intelligent human agents exist in a cooperative social environment that facilitates learning. They learn not only by trialand -error, but also through cooperation by sharing instantaneous information, episodic experience, and learned knowledge. The key investigations of this paper are, "Given the same number of reinforcement learning agents, will cooperative agents outperform independent agents who do not communicate during learning?" and "What is the price for such cooperation?" Using independent agents as a benchmark, cooperative agents are studied in following ways: (1) sharing sensation, (2) sharing episodes, and (3) sharing learned policies. This paper shows that (a) additional sensation from another agent is beneficial if it can be used efficiently, (b) sharing learned policies or episodes among agents speeds up learning at the cost of communication, and (c) for joint tasks, agents engaging in partnership can significantly outperform independent agents although they may learn slo...

It is known that Perceptual Aliasing may significantly diminish the effectiveness of reinforcement learning algorithms [ Whitehead and Ballard, 1991 ] . Perceptual aliasing occurs when multiple situations that are indistinguishable from immediate perceptual input require different responses from the system. For example, if a robot can only see forward, yet the presence of a battery charger behind it determines whether or not it should backup, immediate perception alone is insufficient for determining the most appropriate action. It is problematic since reinforcement algorithms typically learn a control policy from immediate perceptual input to the optimal choice of action. This paper introduces the predictive distinctions approach to compensate for perceptual aliasing caused from incomplete perception of the world. An additional component, a predictive model, is utilized to track aspects of the world that may not be visible at all times. In addition to the control policy, the model mus...

This paper describes a formulation of reinforcement learning that enables learning in noisy, dynamic environemnts such as in the complex concurrent multi-robot learning domain. The methodology involves minimizing the learning space through the use behaviors and conditions, and dealing with the credit assignment problem through shaped reinforcement in the form of heterogeneous reinforcement functions and progress estimators. We experimentally validate the approach on a group of four mobile robots learning a foraging task. 1 Introduction Developing effective methods for real-time learning has been an on-going challenge in autonomous agent research and is being explored in the mobile robot domain. In the last decade, reinforcement learning (RL), a class of approaches in which the agent learns based on reward and punishment it receives from the environment, has become the methodology of choice for learning in a variety of domains, including robotics. In this paper we describe a formulat...

How can artificial neural nets generalize better from fewer examples? In order to generalize successfully, neural network learning methods typically require large training data sets. We introduce a neural network learning method that generalizes rationally from many fewer data points, relying instead on prior knowledge encoded in previously learned neural networks. For example, in robot control learning tasks reported here, previously learned networks that model the effects of robot actions are used to guide subsequent learning of robot control functions. For each observed training example of the target function (e.g. the robot control policy), the learner explains the observed example in terms of its prior knowledge, then analyzes this explanation to infer additional information about the shape, or slope, of the target function. This shape knowledge is used to bias generalization in the learned target function. Results are presented applying this approach to a simulated robot task bas...

We present Utile Suffix Memory, a reinforcement learning algorithm that uses short-term memory to overcome the state aliasing that results from hidden state. By combining the advantages of previous work in instance-based (or "memorybased ") learning and previous work with statistical tests for separating noise from task structure, the method learns quickly, creates only as much memory as needed for the task at hand, and handles noise well. Utile Suffix Memory uses a tree-structured representation, and is related to work on Prediction Suffix Trees [Ron et al., 1994] , Parti-game [Moore, 1993] , G-algorithm [Chapman and Kaelbling, 1991] , and Variable Resolution Dynamic Programming [Moore, 1991] . 1 INTRODUCTION The sensory systems of embedded agents are inherently limited. When a reinforcement learning agent's sensory limitations hide features of the environment from the agent, we say that the agent suffers from hidden state. There are many reasons why important features can be hidden...

Reinforcement learning is a type of unsupervised learning for sequential decision making. Qlearning is probably the best-understood reinforcement learning algorithm. In Q-learning, the agent learns a mapping from states and actions to their utilities. An important assumption of Q-learning is the Markovian environment assumption, meaning that any information needed to determine the optimal actions is reflected in the agent's state representation. Consider an agent whose state representation is based solely on its immediate perceptual sensations. When its sensors are not able to make essential distinctions among world states, the Markov assumption is violated, causing a problem called perceptual aliasing. For example, when facing a closed box, an agent based on its current visual sensation cannot act optimally if the optimal action depends on the contents of the box. There are two basic approaches to addressing this problem--- using more sensors or using history to figure out the curren...