Results 1  10
of
311
Reinforcement learning: a survey
 Journal of Artificial Intelligence Research
, 1996
"... This paper surveys the field of reinforcement learning from a computerscience 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 ..."
Abstract

Cited by 1690 (26 self)
 Add to MetaCart
(Show Context)
This paper surveys the field of reinforcement learning from a computerscience 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 trialanderror 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.
Prioritized sweeping: Reinforcement learning with less data and less time
 Machine Learning
, 1993
"... We present a new algorithm, Prioritized Sweeping, for e cient prediction and control of stochastic Markov systems. Incremental learning methods such asTemporal Di erencing and Qlearning have fast real time performance. Classical methods are slower, but more accurate, because they make full use of ..."
Abstract

Cited by 379 (5 self)
 Add to MetaCart
(Show Context)
We present a new algorithm, Prioritized Sweeping, for e cient prediction and control of stochastic Markov systems. Incremental learning methods such asTemporal Di erencing and Qlearning 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 statespace. We compare Prioritized Sweeping with other reinforcement learning schemes for a number of di erent stochastic optimal control problems. It successfully solves large statespace real time problems with which other methods have di culty. 1 1
Bayesian Experimental Design: A Review
 Statistical Science
, 1995
"... This paper reviews the literature on Bayesian experimental design, both for linear and nonlinear models. A unified view of the topic is presented by putting experimental design in a decision theoretic framework. This framework justifies many optimality criteria, and opens new possibilities. Various ..."
Abstract

Cited by 310 (1 self)
 Add to MetaCart
This paper reviews the literature on Bayesian experimental design, both for linear and nonlinear models. A unified view of the topic is presented by putting experimental design in a decision theoretic framework. This framework justifies many optimality criteria, and opens new possibilities. Various design criteria become part of a single, coherent approach.
Learning to coordinate behaviors
 In Proceedings of AAAI90
, 1990
"... We describe an algorithm which allows a behaviorbased robot to learn on the basis of positive and negative feedback when to activate its behaviors. In accordance with the philosophy of behaviorbased robots, the algorithm is completely distributed: each of the behaviors independently tries to find ..."
Abstract

Cited by 231 (3 self)
 Add to MetaCart
We describe an algorithm which allows a behaviorbased robot to learn on the basis of positive and negative feedback when to activate its behaviors. In accordance with the philosophy of behaviorbased robots, the algorithm is completely distributed: each of the behaviors independently tries to find out (i) whether it is relevant (ie. whether it is at all correlated to positive feedback) and (ii) what the conditions are under which it becomes reliable (i.e. the conditions under which it maximizes the probability of receiving positive feedback and minimizes the probability of receiving negative feedback). The algorithm has been tested successfully on an autonomous 6legged robot which had to learn how to coordinate its legs so as to walk forward. Situation of the Problem Since 1985, the MIT Mobile Robot group has advocated a radically different architecture for autonomous intelligent agents (Brooks, 1986). Instead of decomposing the architecture into functional modules, such as perception, modeling, and planning (figure 1), the architecture is decomposed into taskachieving modules, also called behaviors (figure 2). This novel approach has already demonstrated to be very successful and similar approaches have become more
A ContextualBandit Approach to Personalized News Article Recommendation
"... Personalized web services strive to adapt their services (advertisements, news articles, etc.) to individual users by making use of both content and user information. Despite a few recent advances, this problem remains challenging for at least two reasons. First, web service is featured with dynamic ..."
Abstract

Cited by 170 (16 self)
 Add to MetaCart
(Show Context)
Personalized web services strive to adapt their services (advertisements, news articles, etc.) to individual users by making use of both content and user information. Despite a few recent advances, this problem remains challenging for at least two reasons. First, web service is featured with dynamically changing pools of content, rendering traditional collaborative filtering methods inapplicable. Second, the scale of most web services of practical interest calls for solutions that are both fast in learning and computation. In this work, we model personalized recommendation of news articles as a contextual bandit problem, a principled approach in which a learning algorithm sequentially selects articles to serve users based on contextual information about the users and articles, while simultaneously adapting its articleselection strategy based on userclick feedback to maximize total user clicks. The contributions of this work are threefold. First, we propose a new, general contextual bandit algorithm that is computationally efficient and well motivated from learning theory. Second, we argue that any bandit algorithm can be reliably evaluated offline using previously recorded random traffic. Finally, using this offline evaluation method, we successfully applied our new algorithm to a Yahoo! Front Page Today Module dataset containing over 33 million events. Results showed a 12.5 % click lift compared to a standard contextfree bandit algorithm, and the advantage becomes even greater when data gets more scarce.
Bayesian Qlearning
 In AAAI/IAAI
, 1998
"... A central problem in learning in complex environments is balancing exploration of untested actions against exploitation of actions that are known to be good. The benefit of exploration can be estimated using the classical notion of Value of Information the expected improvement in future decision ..."
Abstract

Cited by 144 (1 self)
 Add to MetaCart
(Show Context)
A central problem in learning in complex environments is balancing exploration of untested actions against exploitation of actions that are known to be good. The benefit of exploration can be estimated using the classical notion of Value of Information the expected improvement in future decision quality that might arise from the information acquired by exploration. Estimating this quantity requires an assessment of the agent's uncertainty about its current value estimates for states. In this paper, we adopt a Bayesian approach to maintaining this uncertain information. We extend Watkins' Qlearning by maintaining and propagating probability distributions over the Qvalues. These distributions are used to compute a myopic approximation to the value of information for each action and hence to select the action that best balances exploration and exploitation. We establish the convergence properties of our algorithm and show experimentally that it can exhibit substantial improvements o...
Solving very large weakly coupled Markov decision processes
 In Proceedings of the Fifteenth National Conference on Artificial Intelligence
, 1998
"... We present a technique for computing approximately optimal solutions to stochastic resource allocation problems modeled as Markov decision processes (MDPs). We exploit two key properties to avoid explicitly enumerating the very large state and action spaces associated with these problems. First, the ..."
Abstract

Cited by 103 (13 self)
 Add to MetaCart
(Show Context)
We present a technique for computing approximately optimal solutions to stochastic resource allocation problems modeled as Markov decision processes (MDPs). We exploit two key properties to avoid explicitly enumerating the very large state and action spaces associated with these problems. First, the problems are composed of multiple tasks whose utilities are independent. Second, the actions taken with respect to (or resources allocated to) a task do not influence the status of any other task. We can therefore view each task as an MDP. However, these MDPs are weakly coupled by resource constraints: actions selected for one MDP restrict the actions available to others. We describe heuristic techniques for dealing with several classes of constraints that use the solutions for individual MDPs to construct an approximate global solution. We demonstrate this technique on problems involving thousandsof tasks, approximating the solution to problems that are far beyond the reach of standard methods. 1
Stochastic linear optimization under bandit feedback
 In submission
, 2008
"... In the classical stochastic karmed bandit problem, in each of a sequence of T rounds, a decision maker chooses one of k arms and incurs a cost chosen from an unknown distribution associated with that arm. The goal is to minimize regret, defined as the difference between the cost incurred by the alg ..."
Abstract

Cited by 98 (8 self)
 Add to MetaCart
(Show Context)
In the classical stochastic karmed bandit problem, in each of a sequence of T rounds, a decision maker chooses one of k arms and incurs a cost chosen from an unknown distribution associated with that arm. The goal is to minimize regret, defined as the difference between the cost incurred by the algorithm and the optimal cost. In the linear optimization version of this problem (first considered by Auer [2002]), we view the arms as vectors in Rn, and require that the costs be linear functions of the chosen vector. As before, it is assumed that the cost functions are sampled independently from an unknown distribution. In this setting, the goal is to find algorithms whose running time and regret behave well as functions of the number of rounds T and the dimensionality n (rather than the number of arms, k, which may be exponential in n or even infinite). We give a nearly complete characterization of this problem in terms of both upper and lower bounds for the regret. In certain special cases (such as when the decision region is a polytope), the regret is polylog(T). In general though, the optimal regret is Θ ∗ ( √ T) — our lower bounds rule out the possibility of obtaining polylog(T) rates in general. We present two variants of an algorithm based on the idea of “upper confidence bounds. ” The first, due to Auer [2002], but not fully analyzed, obtains regret whose dependence on n and T are both essentially optimal, but which may be computationally intractable when the decision set is a polytope. The second version can be efficiently implemented when the decision set is a polytope (given as an intersection √ of halfspaces), but gives up a factor of n in the regret bound. Our results also extend to the setting where the set of allowed decisions may change over time.
MultiArmed Bandits in Metric Spaces
 STOC'08
, 2008
"... In a multiarmed bandit problem, an online algorithm chooses from a set of strategies in a sequence of n trials so as to maximize the total payoff of the chosen strategies. While the performance of bandit algorithms with a small finite strategy set is quite well understood, bandit problems with larg ..."
Abstract

Cited by 92 (11 self)
 Add to MetaCart
In a multiarmed bandit problem, an online algorithm chooses from a set of strategies in a sequence of n trials so as to maximize the total payoff of the chosen strategies. While the performance of bandit algorithms with a small finite strategy set is quite well understood, bandit problems with large strategy sets are still a topic of very active investigation, motivated by practical applications such as online auctions and web advertisement. The goal of such research is to identify broad and natural classes of strategy sets and payoff functions which enable the design of efficient solutions. In this work we study a very general setting for the multiarmed bandit problem in which the strategies form a metric space, and the payoff function satisfies a Lipschitz condition with respect to the metric. We refer to this problem as the Lipschitz MAB problem. We present a complete solution for the multiarmed problem in this setting. That is, for every metric space (L, X) we define an isometry invariant MaxMinCOV(X) which bounds from below the performance of Lipschitz MAB algorithms for X, and we present an algorithm which comes arbitrarily close to meeting this bound. Furthermore, our technique gives even better results for benign payoff functions.
Shifting Inductive Bias with SuccessStory Algorithm, Adaptive Levin Search, and Incremental SelfImprovement
 MACHINE LEARNING
, 1997
"... We study task sequences that allow for speeding up the learner's average reward intake through appropriate shifts of inductive bias (changes of the learner's policy). To evaluate longterm effects of bias shifts setting the stage for later bias shifts we use the "successstory algori ..."
Abstract

Cited by 74 (33 self)
 Add to MetaCart
We study task sequences that allow for speeding up the learner's average reward intake through appropriate shifts of inductive bias (changes of the learner's policy). To evaluate longterm effects of bias shifts setting the stage for later bias shifts we use the "successstory algorithm" (SSA). SSA is occasionally called at times that may depend on the policy itself. It uses backtracking to undo those bias shifts that have not been empirically observed to trigger longterm reward accelerations (measured up until the current SSA call). Bias shifts that survive SSA represent a lifelong success history. Until the next SSA call, they are considered useful and build the basis for additional bias shifts. SSA allows for plugging in a wide variety of learning algorithms. We plug in (1) a novel, adaptive extension of Levin search and (2) a method for embedding the learner's policy modification strategy within the policy itself (incremental selfimprovement). Our inductive transfer case studies...