Abstract
Learning Classifier Systems for Understanding Patterns in Data
Highlights
In the field of data-mining, symbolic techniques, e.g. the Learning classifier systems (LCSs) have produced optimal solutions, which are expected to contain informative patterns
This supports the hypothesis of natural solution and shows these visualization methods are useful in assessing the performance of Learning Classifier Systems (LCSs)’ compaction algorithms, i.e. evaluate the capability of compaction algorithms in ascertaining the optimal rules by visualizing patterns in a compacted model
The proposed visualization techniques can reduce this prejudice since Action-based Feature Importance Map (AFIM) and average Feature Value Map (AFVM) can precisely translate the underlying patterns in the LCSs’ optimal solutions to human-discernable graphs
Summary
In the field of data-mining, symbolic techniques, e.g. the Learning classifier systems (LCSs) have produced optimal solutions, which are expected to contain informative patterns Visualizing these patterns can improve the understanding of the ground truth of the explored domain. Butz et al proposed that LCSs are designed to evolve rulesets that are characterized by completeness, correctness, minimality, and none overlapping to represent the explored domain [18] Such rulesets are considered as the optimal results of LCSs and are termed optimal ruleset [O], which are assumed to contain interpretable patterns that can reflect the underlying nature of the addressed problem (see Section 3.4.3). The training parameters setting for this project are presented
Sign-in/Register to view highlights & summary 
Submitted Version (
Free)
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call