

...problem and derive
a new algorithm that adapts decision tree learning techniques to state abstraction. the proof of performance is supported...trials, agent’s performance, size of the learned trees, and computation time of the algorithm.
keywords: state abstraction, hierarchical reinforcement learning
...
Published in 2006.

...problems, such as robot navigation using highresolution sensors. selforganizing distinctive state abstraction (soda) is a new, generic...carry it between dis tinctive states based on those features. a soda agent first uses a self organizing feature map...

...we study automatic generation of options using common subsequences derived from the state transition histories collected as learning progresses. the standard qlearning algorithm is extended to use generated options transparently, and effectiveness of the method is demonstrated in dietterich...

abstract—traditionally, performance has been the most im portant metrics when evaluating a system. however, in the last decades,industry,and,academia...show,that m5p (a decision tree algorithm) is the best option to model,the behavior,of the system,under,the random,injection of memory,leaks...

...semimarkov decision problem (smdp). the algorithm uses a learned tree based discretization of the state space as an abstract state descrip tion and both user supplied and autogenerated policies as temporally abstract actions. it uses a generative model of the world to sample the transition function for the abstract smdp defined by those state...

...this paper we investigate in the possible application of parameterized verification...identify a class of infinitestate abstract models that preserve the main features of the semantics of concurrent java. we...

...s abstract— action rules, interventions, and eaction rules are examples of knowledge discovery tools for reclassification of objects and for defining actionability as a partially...rule construction. they differ in the splitting criterion applied at every node of a generated tree. in this paper, actiontree algorithm is utilized. we focus on...

...we illustrate the use of logic programming techniques for finite model checking of ctl formulae. we present a technique for infi nite state model checking of safety properties based upon logic program specialisation and analysis techniques. the power of the approach is il lustrated on several examples. for that, the e!ci ent tools logen...

...have been developing a theory for the generic representation of 2d shape, where structural de scriptions are derived from the shocks (singularities) of a curve...to a unique rooted shock tree. we introduce a novel tree matching algorithm which finds the best set of corresponding nodes...