Autoencoding With a Classifier System

Preen, Richard J.; Wilson, Stewart W.; Bull, Larry

doi:10.1109/tevc.2021.3079320

Cited by 12 publications

(13 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following [10], here XCSF maintains a population of classifiers where each cl.C and cl.P is a separate neural network. A population set [P ] of N = 500 classifiers are initialised randomly and undertake Lamarckian learning.…”

Section: A Overviewmentioning

confidence: 99%

“…Following [10], crossover is omitted from the EA and a layer-specific self-adaptive mutation scheme is used. That is, the mutation rates are locally evolving entities, continuously adapting throughout the search process.…”

Section: B Evolutionary Operatorsmentioning

confidence: 99%

“…Each sub-solution is activated only for the subset of instances that its condition permits. Error is therefore allocated directly to the components responsible, potentially enabling faster convergence and reduced computational cost [10].…”

mentioning

confidence: 99%

“…Shallow neural networks have been widely explored within LCS for supervised and reinforcement learning tasks [11]- [19]. More recently, they have been used in the unsupervised learning of an ensemble of heterogeneous autoencoders [10]. Spiking neural networks [20] and cyclic graphs [21], [22], which are capable of deep architectures have also been explored, although these do not include the ability for refinement with local search.…”

mentioning

confidence: 99%

“…We explore the performance of fully-connected and convolutional neural networks where the number of units as well as the connectivity evolves, i.e., deep heterogeneous niched networks may emerge. Furthermore, as in [10], we employ a self-adaptive scheme wherein the rate of mutation is continually adapted throughout the search process. Each layer adapts to a localised rate of gradient descent.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Deep Learning with a Classifier System: Initial Results

Preen,

Bull

2021

Preprint

Self Cite

View full text Add to dashboard Cite

This article presents the first results from using a learning classifier system capable of performing adaptive computation with deep neural networks. Individual classifiers within the population are composed of two neural networks. The first acts as a gating or guarding component, which enables the conditional computation of an associated deep neural network on a per instance basis. Self-adaptive mutation is applied upon reproduction and prediction networks are refined with stochastic gradient descent during lifetime learning. The use of fullyconnected and convolutional layers are evaluated on handwritten digit recognition tasks where evolution adapts (i) the gradient descent learning rate applied to each layer (ii) the number of units within each layer, i.e., the number of fully-connected neurons and the number of convolutional kernel filters (iii) the connectivity of each layer, i.e., whether each weight is active (iv) the weight magnitudes, enabling escape from local optima. The system automatically reduces the number of weights and units while maintaining performance after achieving a maximum prediction error.

show abstract

Section: A Overviewmentioning

confidence: 99%

Section: B Evolutionary Operatorsmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Deep Learning with a Classifier System: Initial Results

Preen,

Bull

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Deep Reinforcement Learning with a Classifier System – First Steps

Schönberner

Tomforde²

2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Internalizing Knowledge for Anticipatory Classifier Systems in Discretized Real-Valued Environments

Kozłowski

Unold

2022

IEEE Access

View full text Add to dashboard Cite

Real-valued environments are challenging for learning systems because of a significant increase in the input space size of the problem. This work demonstrates that Anticipatory Learning Classifier Systems (ALCS) can successfully build sets of conditional rules foreseeing the consequences of executed actions. Three major classes of Learning Classifier Systems -Anticipatory Classifier System (ACS), ACS2, Yet Another Classifier System (YACS), alongside the traditional Dyna-Q algorithm implementations were adapted to handle real-valued input signal discretization. Aspects like the ability to capture all possible interactions, model generalization capabilities, size of the solution of relative execution times were compared in four different problems using probabilistic modelling, providing unbiased judgments. Results proved that the examined ALCS are capable of solving selected problems. Despite increased input size, all possible environmental transitions were learned latently, without obtaining any explicit incentives. Such an internal representation provides a more compact solution representation and can optimize learning speed further by executing imaginary environmental interactions or performing action planning for a new set of potential problems.

show abstract

Autoencoding With a Classifier System

Cited by 12 publications

References 39 publications

Deep Learning with a Classifier System: Initial Results

Deep Learning with a Classifier System: Initial Results

Deep Reinforcement Learning with a Classifier System – First Steps

Internalizing Knowledge for Anticipatory Classifier Systems in Discretized Real-Valued Environments

Contact Info

Product

Resources

About