Analysing Learning Classifier Systems in Reactive and Non-reactive Robotic Tasks

Moioli, Renan Cipriano; Vargas, Patrícia A.; Zuben, Fernando J. Von

doi:10.1007/978-3-540-88138-4_17

Cited by 4 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More recently, Butz and Herbort (2008) demonstrated an XCSF-derivative to control a robot arm using a real interval representation. Recent research by Moioli et al (2007) compared attempts at T-maze navigation in both simulated and real robotics tasks, and presented results that reinforce the idea that TCS is a valid method for dealing with long-action chains that are present in many robotic environments. Similarly to Webb et al (2003), the authors added a memory register and tested the system on a non-reactive robotics task.…”

Section: Lcs Robotics Researchmentioning

confidence: 67%

A Spiking Neural Learning Classifier System

Howard,

Bull,

Lanzi

2012

Preprint

View full text Add to dashboard Cite

Learning Classifier Systems (LCS) are populationbased 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. We employ a constructivist model of growth of both neurons and dendrites that realise flexible learning by evolving structures of sufficient complexity to solve a well-known problem involving continuous, real-valued inputs. Additionally, we extend the system to enable temporal state decomposition. By allowing our LCS to chain together sequences of heterogeneous actions into macro-actions, it is shown to perform optimally in a problem where traditional methods can fail to find a solution in a reasonable amount of time. Our final system is tested on a simulated robotics platform. Keywords Learning Classifier SystemsMany real-world processes encompass some temporal element that describes or affects the state of a given system through time. Artificial Intelligence ubiquitously utilises abstractions of real-world or natural systems, however many representations opt to remove these temporal facets during the abstraction process. Missing out on underlying temporal patterns in a problem potentially denies a system access to a

show abstract

Section: Lcs Robotics Researchmentioning

confidence: 67%

A Spiking Neural Learning Classifier System

Howard,

Bull,

Lanzi

2012

Preprint

View full text Add to dashboard Cite

show abstract

“…Further examples of TCS robot control include a demonstration of the ability to disambiguate between aliased states using an internal memory register [49]. More recent research [36] compares results of a T-maze navigation task, again using a memory register to handle aliased states. Results show strong performance in both real and simulated robotics tasks.…”

Section: Learning Classifier System Roboticsmentioning

confidence: 99%

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

Howard¹,

Bull

Lanzi

2015

Neural Process Lett

View full text Add to dashboard Cite

Learning Classifier Systems (LCS) are population-based reinforcement learners that were originally designed to model various cognitive phenomena. This paper presents an explicitly cognitive LCS by using spiking neural networks as classifiers, providing each classifier with a measure of temporal dynamism. 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. The spiking classifiers are coupled with a temporally-sensitive reinforcement learning algorithm, which allows the system to perform temporal state decomposition by appropriately rewarding "macro-actions", created by chaining together multiple atomic actions. The combination of temporal reinforcement learning and neural information processing is shown to outperform benchmark neural classifier systems, and successfully solve a robotic navigation task.

show abstract

“…This approach was first used by Holland in the original LCS framework (Holland and Reitman 1977) with the use of an internal message list, but showed only limited success on non-Markov problems (Smith 1994). And afterward, a different form of the explicit internal state solution was adopted in ZCSM (Cliff and Ross 1994) and XCSMH (Lanzi 1998a;Lanzi and Wilson 2000;Moioli et al 2008) by using memory register.…”

Section: T T F T T T T T T T T T T T T Tmentioning

confidence: 99%

Learning classifier systems with memory condition to solve non-Markov problems

Zang

Wang

2014

Soft Comput

View full text Add to dashboard Cite

In the family of Learning Classifier Systems, the classifier system XCS has been successfully used for many applications. However, the standard XCS has no memory mechanism and can only learn optimal policy in Markov environments, where the optimal action is determined solely by the state of current sensory input. In practice, most environments are partially observable environments on agent's sensation, which are also known as non-Markov environments. Within these environments, XCS either fails, or only develops a suboptimal policy, since it has no memory. In this work, we develop a new classifier system based on XCS to tackle this problem. It adds an internal message list to XCS as the memory list to record input sensation history, and extends a small number of classifiers with memory conditions. The classifier's memory condition, as a foothold to disambiguate non-Markov states, is used to sense a specified element in the memory list. Besides, a detection method is employed to recognize non-Markov states in environments, to avoid these states controlling over classifiers' memory conditions. Furthermore, four sets of different complex maze environments have been tested by the proposed method. Experimental results show that our system is one of the best techniques to solve partially observable environments, compared with some well-known classifier systems proposed for these environments.

show abstract

Analysing Learning Classifier Systems in Reactive and Non-reactive Robotic Tasks

Cited by 4 publications

References 25 publications

A Spiking Neural Learning Classifier System

A Spiking Neural Learning Classifier System

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

Learning classifier systems with memory condition to solve non-Markov problems

Contact Info

Product

Resources

About