Several recent efforts in the field of reinforcement learning have focused attention on the importance of regularization, but the techniques for incorporating regularization into reinforcement learning algorithms, and the effects of these changes upon the convergence of these algorithms, are ongoing areas of research. In particular, little has been written about the use of regularization in onl...

ing Reusable Cases from Reinforcement Learning Andreas von Hessling and Ashok K. Goel College of Computing Georgia Institute of Technology Atlanta, GA 30318 {avh, goel}@cc.gatech.edu Abstract. Reinforcement Learning is a popular technique for gameplaying because it can learn an optimal policy for sequential decision problems in which the outcome (or reward) is delayed. However, Reinforcement Le...

The aim of this work is to combine three successful AI techniques –Reinforcement Learning (RL), Heuristics Search and Case Based Reasoning (CBR)– creating a new algorithm that allows the use of cases in a case base as heuristics to speed up Reinforcement Learning algorithms. This approach, called Case Based Heuristically Accelerated Reinforcement Learning (CB-HARL), builds upon an emerging tech...

The DRL (Delayed Reinforcement Learning) problem is classical in Reinforcement Learning theory. There were several agent architectures solving that problem including some connectionist architectures. This work describes an early connectionist agent architecture, the CAA architecture, that solved the problem using the concept of emotion it its learning rule. The architecture is compared to anoth...

Learning Classifier Systems use reinforcement learning, evolutionary computing and/or heuristics to develop adaptive systems. This paper extends the ZCS Learning Classifier System to improve its internal modelling capabilities. Initially, results are presented which show performance in a traditional reinforcement learning task incorporating lookahead within the rule structure. Then a mechanism ...

This paper examines the suitability of Lego Mindstorms robotic kits as a platform for teaching the concepts of reinforcement learning. The reinforcement learning algorithm Sarsa was implemented on board an autonomous Mindstorms robot, and applied to two learning tasks. The reasons behind the differing results obtained on these two tasks are discussed, and several issues related to the suitab...

Reinforcement learning is a machine learning area that studies which actions an agent can take in order to optimize a cumulative reward function. Recently, a new class of reinforcement learning algorithms with multiple, possibly conflicting, reward functions was proposed. We call this class of algorithms the multi-objective reinforcement learning (MORL) paradigm. We give an overview on multi-ob...

Reinforcement learning is a general formalism for sequential decision-making, with recent algorithm development focusing on function approximation to handle large state spaces and high-dimensional, high-velocity (sensor) data. The success of function approximators, however, hinges on the quality of the data representation. In this work, we explore representation learning within batch reinforcem...

A key goal of AI is to create lifelong learning agents that can leverage prior experience to improve performance on later tasks. In reinforcement learning problems, one way to summarize prior experience for future use is through options, which are behaviorally extended actions (subpolicies) for how to behave. Options can then be used to potentially accelerate learning in new reinforcement learn...

The search for interpretable reinforcement learning policies is of high academic and industrial interest. Especially for industrial systems, domain experts are more likely to deploy autonomously learned controllers if they are understandable and convenient to evaluate. Basic algebraic equations are supposed to meet these requirements, as long as they are restricted to an adequate complexity. He...