Is k Nearest Neighbours Regression Better Than GP?

Leonardo Vanneschi, Mauro Castelli, Luca Manzoni, Sara Silva, Leonardo Trujillo

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Citation (Scopus)


This work starts from the empirical observation that k nearest neighbours (KNN) consistently outperforms state-of-the-art techniques for regression, including geometric semantic genetic programming (GSGP). However, KNN is a memorization, and not a learning, method, i.e. it evaluates unseen data on the basis of training observations, and not by running a learned model. This paper takes a first step towards the objective of defining a learning method able to equal KNN, by defining a new semantic mutation, called random vectors-based mutation (RVM). GP using RVM, called RVMGP, obtains results that are comparable to KNN, but still needs training data to evaluate unseen instances. A comparative analysis sheds some light on the reason why RVMGP outperforms GSGP, revealing that RVMGP is able to explore the semantic space more uniformly. This finding opens a question for the future: is it possible to define a new genetic operator, that explores the semantic space as uniformly as RVM does, but that still allows us to evaluate unseen instances without using training data?

Original languageEnglish
Title of host publicationGenetic Programming - 23rd European Conference, EuroGP 2020, Held as Part of EvoStar 2020, Proceedings
EditorsTing Hu, Nuno Lourenço, Eric Medvet, Federico Divina
Number of pages18
ISBN (Print)9783030440930
Publication statusPublished - 9 Apr 2020
Event23rd European Conference on Genetic Programming, EuroGP 2020, held as part of EvoStar 2020 - Seville, Spain
Duration: 15 Apr 202017 Apr 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12101 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference23rd European Conference on Genetic Programming, EuroGP 2020, held as part of EvoStar 2020


Dive into the research topics of 'Is k Nearest Neighbours Regression Better Than GP?'. Together they form a unique fingerprint.

Cite this