Classifier Fitness Based on Accuracy

@article{Wilson1995ClassifierFB,
  title={Classifier Fitness Based on Accuracy},
  author={Stewart W. Wilson},
  journal={Evolutionary Computation},
  year={1995},
  volume={3},
  pages={149-175}
}
  • Stewart W. Wilson
  • Published 1 June 1995
  • Mathematics, Computer Science
  • Evolutionary Computation
In many classifier systems, the classifier strength parameter serves as a predictor of future payoff and as the classifier's fitness for the genetic algorithm. [...] Key Result Besides introducing a new direction for classifier system research, these properties of XCS make it suitable for a wide range of reinforcement learning situations where generalization over states is desirable.Expand
Classifier Systems -- Accuracy-Based Fitness Allows . . .
Traditionally within classifier systems the abili ty of a classifier to obtain reward (as measured by its strength) indicates the fitness of the classifier within the rule population. However Wilson
A simple accuracy-based learning classifier system
TLDR
This paper presents a simple accuracy-based learning classifier system with which to explore aspects of accuracy- based fitness in general, and describes and modelled, before being implemented and tested on the multiplexer task.
Accuracy-based fitness allows similar performance to humans in static and dynamic classification environments
TLDR
This paper presents experiments with two different classifier systems: Newboole (Bonelli et al. 1990) and XCS (Wilson 1995), which demonstrate qualitative matches to data from perceptual category learning in humans and the different methods of fitness evaluation of classifiers alter the knowledge the systems learn and maintain.
Using Raw Accuracy to Estimate Classifier Fitness in XCS
TLDR
In XCS, classifier fitness does not provide information about the problem solution, but rather an indication of the classifier relevance in the encountered situations, so it is not generally possible to tell whether a classifier with a high fitness is accurate or not, just looking at the fitness.
Strength or Accuracy? Fitness Calculation in Learning Classifier Systems
  • T. Kovacs
  • Computer Science
    Learning Classifier Systems
  • 1999
TLDR
It is concluded that XCS's accuracy-based fitness appears to have a number of significant advantages over traditional strength-based Fitness, including an apparent advantage in addressing the explore/exploit problem.
A comparison of relative accuracy and raw accuracy in XCS
  • P. Lanzi
  • Computer Science, Materials Science
    The 2003 Congress on Evolutionary Computation, 2003. CEC '03.
  • 2003
TLDR
A modification of Wilson's original definition in which classifier fitness is measured as the absolute (raw) accuracy of classifier prediction is introduced, and Wilson's relative accuracy and raw accuracy on a number of problems are compared.
Simple Markov Models of the Genetic Algorithm in Classifier Systems: Accuracy-Based Fitness
  • L. Bull
  • Mathematics, Computer Science
    IWLCS
  • 2000
TLDR
A simple Markov model of the algorithm in Michigan-style Classifier Systems, allowing comparison between the two forms of rule utility measure, and the previously discussed benefits of accuracy over prediction are clearly shown with regard to overgeneral rules.
Improving genetic search in XCS-based classifier systems through understanding the evolvability of classifier rules
TLDR
A method to trace the evolution of classifier rules generated in an XCS-based classifier system using the concept of a family tree, termed parent-tree, for each individual classifier rule generated in the system during training, which describes the whole generational process for that classifier.
On lookahead and latent learning in simple LCS
  • L. Bull
  • Computer Science
    GECCO '07
  • 2007
TLDR
This paper presents a simple but effective learning classifier system of this last type, using payoff-based fitness, with the aim of enabling the exploration of their basic principles, i.e., in isolation from the many other mechanisms they usually contain.
On Lookahead and Latent Learning in Simple LCS
TLDR
This paper presents a simple but effective learning classifier system of this last type, using payoff-based fitness, with the aim of enabling the exploration of their basic principles, i.e., in isolation from the many other mechanisms they usually contain.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 50 REFERENCES
Credit Assignment for Pole Balancing with Learning Classifier Systems. In
TLDR
A classifier system, XCS, is investigated, in which each classifier maintains a prediction of expected payoff, but the classifier’s fitness is given by a measure of the prediction's accuracy, which makes it suitable for a wide range of reinforcement learning situations where generalization over states is desirable.
COGNITIVE SYSTEMS BASED ON ADAPTIVE ALGORITHMS1
The type of cognitive system (CS) studied here has four basic parts: (1) a set of interacting elementary productions, called lassifiers, (2) a performance algorithm that directs the action of the
Implicit Niching in a Learning Classifier System: Nature's Way
TLDR
This work isolates one crucial subfunction in the LCS learning algorithm: covering through niching, and can bring results to bear in understanding the fundamental type of cooperation (a.k.a. weak cooperation) that an LCS must promote.
Cognitive systems based on adaptive algorithms
TLDR
The type of cognitive system studied here has a set of interacting elementary productions, called classifiers, and a performance algorithm that directs the action of the system in the environment and modifies the set of classifiers so that variants of good classifiers persist and new, potentially better ones are created in a provably efficient manner.
Evolutionary learning of general fuzzy rules with biased evaluation functions: competition and cooperation
  • Andrea Bonarini
  • Computer Science
    Proceedings of the First IEEE Conference on Evolutionary Computation. IEEE World Congress on Computational Intelligence
  • 1994
TLDR
This work discusses issues, and presents a solution, implemented in ELF (evolutionary learning of fuzzy rules), which was successfully applied to develop autonomous agents, and other fuzzy controlled devices.
ZCS: A Zeroth Level Classifier System
A basic classifier system, ZCS, is presented that keeps much of Holland's original framework but simplifies it to increase understandability and performance. ZCS's relation to Q-learning is brought
Default hierarchy formation and memory exploitation in learning classifier systems
TLDR
A class of problems that isolate memory exploitation from other aspects of LCS behavior is developed, showing that an LCS can form rule sets that exploit memory and does not form optimal rule sets because of a limitation in its allocation of credit scheme.
Hierarchical Credit Allocation in a Classifier System
TLDR
This work suggests an alternative form for the algorithm and the system's operating principles designed to induce behavioral hierarchies in which modularity of the hierarchy would keep all bucket-brigade chains short, thus more reinforceable and more rapidly learned, but overall action sequences could be long.
Intelligent behavior as an adaptation to the task environment ; Part II.
TLDR
This dissertation argues that examining more closely the way animate systems cope with real-world environments can provide valuable insights about the structural requirements for intelligent behavior.
A Critical Review of Classifier Systems
TLDR
This work presents a kind of rule-based system with general mechanisms for processing rules in parallel, for adaptive generation of new rules, and for testing the effectiveness of existing rules.
...
1
2
3
4
5
...