Filters








27 Hits in 1.4 sec

Autoencoding with a Classifier System [article]

Richard J. Preen and Stewart W. Wilson and Larry Bull
2021 arXiv   pre-print
LCS perform conditional computation through the use of a population of individual gating/guarding components, each associated with a local approximation.  ...  This article explores the use of an LCS to adaptively decompose the input domain into a collection of small autoencoders where local solutions of different complexity may emerge.  ...  Recently, [43] have investigated an LCS where the EA performs feature selection using bitstring conditions and a selection of convolutional neural network actions are used.  ... 
arXiv:1910.10579v7 fatcat:kixxs32gtbcpncscbfjtfiptta

On the effects of node duplication and connection-oriented constructivism in neural XCSF

Gerard David Howard, Larry Bull
2008 Proceedings of the 2008 GECCO conference companion on Genetic and evolutionary computation - GECCO '08  
The system uses a rule structure in which each is represented by an artificial neural network.  ...  This paper presents the use of constructivism-inspired mechanisms within a neural learning classifier system which exploits parameter self-adaptation as an approach to realize such behavior.  ...  In particular, we explore the success of extensions to the XCSF-based neural LCS, N-XCSF [11] , including node duplication, and a connection-oriented neural constructivism scheme, on a real-valued version  ... 
doi:10.1145/1388969.1389010 dblp:conf/gecco/HowardB08 fatcat:2uhdczusnbcjridambe7fcmgqi

Fuzzy dynamical genetic programming in XCSF

Richard J. Preen, Larry Bull
2011 Proceedings of the 13th annual conference companion on Genetic and evolutionary computation - GECCO '11  
A number of representation schemes have been presented for use within Learning Classifier Systems, ranging from binary encodings to Neural Networks, and more recently Dynamical Genetic Programming (DGP  ...  This paper presents results from an investigation into using a fuzzy DGP representation within the XCSF Learning Classifier System.  ...  Teller and Veloso's "neural programming" [45] uses a directed graph of connected nodes, each performing an arbitrary function.  ... 
doi:10.1145/2001858.2001952 dblp:conf/gecco/PreenB11 fatcat:dqlr5b6ugnbchpyc3lkpsrkywe

Discrete and fuzzy dynamical genetic programming in the XCSF learning classifier system

Richard J. Preen, Larry Bull
2013 Soft Computing - A Fusion of Foundations, Methodologies and Applications  
A number of representation schemes have been presented for use within learning classifier systems, ranging from binary encodings to neural networks.  ...  It is shown possible to use self-adaptive, open-ended evolution to design an ensemble of such dynamical systems within XCSF to solve a number of well-known test problems.  ...  Neural Programming (NP) [91] uses a directed graph of connected nodes, each performing an arbitrary function.  ... 
doi:10.1007/s00500-013-1044-4 fatcat:wlt6myftjrfehj2sy4s4dmgwmm

A Spiking Neural Learning Classifier System [article]

Gerard Howard and Larry Bull and Pier-Luca Lanzi
2012 arXiv   pre-print
Learning Classifier Systems (LCS) are population-based reinforcement learners used in a wide variety of applications.  ...  This paper presents a LCS where each traditional rule is represented by a spiking neural network, a type of network with dynamic internal state.  ...  Connection Selection Feature selection (FS) is a method of reducing the number of the data inputs to a process by selecting and operating exclusively on a subset of those inputs.  ... 
arXiv:1201.3249v1 fatcat:kimwxfr2vjc5ncmjwyw5wbr4uq

A brief history of learning classifier systems: from CS-1 to XCS and its variants

Larry Bull
2015 Evolutionary Intelligence  
The use of selection within niches of co-active rules is akin to the scheme used in a general class of AIS known as clonal selection algorithms (e.g., CLONALG [De Castro & Von Zuben, 2002] ).  ...  Given the supervised learning-like nature of building anticipations, they have also been learned in a version of XCSF using a neural network to predict the next state (e.g., ).  ... 
doi:10.1007/s12065-015-0125-y fatcat:tjihtlllsrfhtnohjqehxtkrf4

A Brief History of Learning Classifier Systems: From CS-1 to XCS [article]

Larry Bull
2014 arXiv   pre-print
Modern Learning Classifier Systems can be characterized by their use of rule accuracy as the utility metric for the search algorithm(s) discovering useful rules.  ...  This paper gives an historical overview of the evolution of such systems up to XCS, and then some of the subsequent developments of XCS to different types of learning.  ...  The use of selection within niches of co-active rules is akin to the scheme used in a general class of AIS known as clonal selection algorithms (e.g., CLONALG [De Castro & Von Zuben, 2002] ).  ... 
arXiv:1401.3607v2 fatcat:t4qk5a3ljrhgvmjz7xl5swbtzi

Optimality-Based Analysis of XCSF Compaction in Discrete Reinforcement Learning [chapter]

Jordan T. Bishop, Marcus Gallagher
2020 Lecture Notes in Computer Science  
A well-studied LCS architecture is XCSF, which in the RL setting acts as a Q-function approximator.  ...  We apply XCSF to a deterministic and stochastic variant of the FrozenLake8x8 environment from OpenAI Gym, with its performance compared in terms of function approximation error and policy accuracy to the  ...  Compared to other Q-function approximators used in RL (e.g. neural networks), XCSF has the advantage of presenting its knowledge in a piecewise, easily interpretable format that can reduced to a compact  ... 
doi:10.1007/978-3-030-58115-2_33 fatcat:4rfouxsycfdsrjo64gnfo6puum

A Cognitive Architecture Based on a Learning Classifier System with Spiking Classifiers [article]

David Howard, Larry Bull, Pier-Luca Lanzi
2015 arXiv   pre-print
We employ a constructivist model of growth of both neurons and synaptic connections, which permits a Genetic Algorithm (GA) to automatically evolve sufficiently-complex neural structures.  ...  This paper presents an explicitly cognitive LCS by using spiking neural networks as classifiers, providing each classifier with a measure of temporal dynamism.  ...  In this way, an LCS can generate optimal action selection policies in an environment. An LCS therefore consists of three main components: a GA, an RL scheme, and a population of classifiers.  ... 
arXiv:1508.07700v1 fatcat:7loktdgxybe3bmc7sgnpcvjvzm

A Cognitive Architecture Based on a Learning Classifier System with Spiking Classifiers

David Howard, Larry Bull, Pier-Luca Lanzi
2015 Neural Processing Letters  
A cognitive architecture based on a learning classifier system with spiking classifiers.  ...  In this way, an LCS can generate optimal action selection policies in an environment. An LCS therefore consists of three main components: a GA, an RL scheme, and a population of classifiers.  ...  scheme with a temporally sensitive neural classifier representation are enhanced as the amount of temporal information in the environment increases.  ... 
doi:10.1007/s11063-015-9451-4 fatcat:7jfdklzyjzettknqrpvci4erie

Learning local linear Jacobians for flexible and adaptive robot arm control

Patrick O. Stalph, Martin V. Butz
2011 Genetic Programming and Evolvable Machines  
We are interested in a learning approach that can adapt to such changes-be they due to motor or sensory failures, or also due to the flexible extension of the robot body by, for example, the usage of tools  ...  For detailed evaluation purposes, we study the performance of XCSF when learning to control an anthropomorphic seven degrees of freedom arm in simulation.  ...  Open Access This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided  ... 
doi:10.1007/s10710-011-9147-0 fatcat:5gruwdjiy5bmlfkr2wcdpvfpyy

Knowledge Representation in Learning Classifier Systems: A Review [article]

Farzaneh Shoeleh, Mahshid Majd, Ali Hamzeh, Sattar Hashemi
2015 arXiv   pre-print
This component brings insight into how to partition the problem space what in turn seeks prominent role in generalization capacity of the system as a whole.  ...  The current work is an attempt to find a comprehensive and yet elaborate view into the existing knowledge representation techniques in LCS domain in general and XCS in specific.  ...  One of the main advantages of this scheme, namely using neural network based representation, is its ability to be applied on problems with continuous action space.  ... 
arXiv:1506.04002v1 fatcat:mpwk7ga3azbh5akoe3coaydp2q

On-line regression algorithms for learning mechanical models of robots: A survey

Olivier Sigaud, Camille Salaün, Vincent Padois
2011 Robotics and Autonomous Systems  
In this paper, we provide a survey of the corresponding literature with a focus on the methods rather than on the results.  ...  In particular, we provide a unified view of all recent algorithms that outlines their distinctive features and provides a framework for their combination.  ...  XCSF xcsf shares a lot of similarities with lwpr in the structure of the learned models.  ... 
doi:10.1016/j.robot.2011.07.006 fatcat:mi46xi5l7bep5ijfzbdgiv7dqa

Topologically controlled segmentation of 3D magnetic resonance images of the head by using morphological operators

Petr Dokládal, Isabelle Bloch, Michel Couprie, Daniel Ruijters, Raquel Urtasun, Line Garnero
2003 Pattern Recognition  
The morphology-oriented approach combined with an extensive use of topological constraints provides a robust and automatic method requiring minimum user intervention.  ...  The originality of the approach lies in the satisfaction of such constraints and in an e ort towards robustness.  ...  [34] ) are usually based on the following scheme: (1) selection of markers of the objects to extract, (2) reconstruction of boundaries between the marked objects by using the watersheds.  ... 
doi:10.1016/s0031-3203(03)00118-3 fatcat:gamrv74q2bgebgatammk4a4h3u

Learning Classifier Systems: A Complete Introduction, Review, and Roadmap

Ryan J. Urbanowicz, Jason H. Moore
2009 Journal of Artificial Evolution and Applications  
This paper aims to provide an accessible foundation for researchers of different backgrounds interested in selecting or developing their own LCS.  ...  Included is a simple yet thorough introduction, a historical review, and a roadmap of algorithmic components, emphasizing differences in alternative LCS implementations.  ...  Once the match set is established, an action is selected using a simple explore/exploit scheme [22] .  ... 
doi:10.1155/2009/736398 fatcat:dvubeerbl5ez3ed553sn2snofy
« Previous Showing results 1 — 15 out of 27 results