Ogrodje NiaAML za samodejno strojno učenje

magistrsko delo

Luka Pečnik (Author), Iztok Fister (Mentor), Iztok Fister (Co-mentor)

Abstract

V magistrskem delu smo raziskali področje samodejnega strojnega učenja in natančneje metodo za samodejno strojno učenje, imenovano NiaAML. Osredotočili smo se predvsem na iskanje klasifikacijskih cevovodov s pomočjo stohastičnih populacijskih algoritmov po vzorih iz narave. S pomočjo programskega jezika Python in knjižnic, ki jih ponuja, smo razvili istoimensko ogrodje za samodejno strojno učenje NiaAML, namenjeno iskanju in optimizaciji klasifikacijskih cevovodov. V ogrodju smo metodo NiaAML poskusili še izboljšati, nato pa smo primerjali rezultate med originalno in spremenjeno metodo NiaAML.

Keywords

algoritmi po vzorih iz narave;klasifikacijski cevovodi;samodejno strojno učenje;magistrske naloge;

Data

Language:	Slovenian
Year of publishing:	2021
Typology:	2.09 - Master's Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	[L. Pečnik]
UDC:	004.85.021(043.2)
COBISS:	54864387
Views:	547
Downloads:	117
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Automated Machine Learning Framework NiaAML
Secondary abstract:	In this thesis, we researched the field of automatic machine learning and, more precisely, the method for automatic machine learning called NiaAML. We focused mainly on searching for classification pipelines using stochastic population-based nature-inspired algorithms. With the help of the Python programming language and the libraries it offers, we have also developed a framework of the same name for finding and optimizing classification pipelines. We tried to further improve the NiaAML method in the framework, and then compared the results between the original and the modified NiaAML method.
Secondary keywords:	automated machine learning;classification pipelines;nature-inspired algorithms;
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages:	XIII, 62 f.
ID:	12437394