2023

abstract

english

EN Interpretability is the next frontier in machine learning research. In the search for white box models - as opposed to black box models like random forests or neural networks - rule induction algorithms are a logical and promising option since the rules they create can easily be understood by humans. Fuzzy and rough set theories have been applied separately to this archetype, to great success. As both applications involve processing equivalence classes, it is only natural to combine them. The QuickRules algorithm was the first attempt at using fuzzy rough set theory for rule induction. It is based on QuickReduct, a greedy algorithm for building decision reducts. QuickRules already showed an improvement over other rule induction methods, but we can further improve on it by using fuzzy rough sets based on ordered weighted average (OWA) operators, which reduces the impact of outliers, and by removing redundant rules in a pruning step, without decreasing accuracy. However, to evaluate the full potential of a fuzzy rough rule induction algorithm, we need to start from the ground up. A fuzzy indiscernibility relation divides the data space into fuzzy granules, which we could combine from the bottom up or create in a top-down approach, both of which could result in a successful rule set. We also need to decide what kind of rule set we require: what is a minimal covering set in this context, and is that desirable? Moreover, we should figure out how to decide which attributes each rule needs to contain, and how to perform the inference process. Additionally, we should evaluate the effect of using OWA-based fuzzy rough sets on the performance and rule set size of the algorithm. Finally, a study evaluating the effect of the indiscernibility relation should also be performed.

139

https://uibes-my.sharepoint.com/personal/smm900_id_uib_eu/_layouts/15/onedrive.aspx?id=%2Fpersonal%2Fsmm900%5Fid%5Fuib%5Feu%2FDocuments%2FFitxers%20adjunts%2FEUSFLAT%2DAGOP%2D2023%2DBookAbstracts%2Epdf&parent=%2Fpersonal%2Fsmm900%5Fid%5Fuib%5Feu%2FDocuments%2FFitxers%20adjunts&ga=1

EUSFLAT 2023 and AGOP 202313th : Conference of the European Society for Fuzzy Logic, 12th International Summer School on Aggregation Operators, September 4–8, 2023, Palma, Spain : Book of Abstracts

13th Conference of the European Society for Fuzzy Logic and Technology, EUSFLAT 2023, and 12th International Summer School on Aggregation Operators, AGOP 2023, 4-8.09.2023, Palma de Mallorca, Spain

CC BY (attribution alone)

publisher's website

final author's version

at the time of publication