Višerazredna klasifikacija čvorova grafa na osnovi naučenih značajki

master's thesis

• Title (english): Multiclass Classification of Graph Nodes Based on Learned Features
• Author: Mislav Požek
• Mentor: Marin Šilić (mentor)
• Committee member: Marin Šilić (predsjednik povjerenstva)
• Committee member: Dejan Škvorc (član povjerenstva)
• Committee member: Goran Delač (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-02-21, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu proučene su i uspoređene tri suvremene metode otkrivanja značajki čvorova grafova; DeepWalk, LINE, Node2Vec.
Korištene su tri vrste čestih klasifikatora: model logističke regresije, grupiranje najbližim centroidom i stroj potpornih vektora.
Primjenom pronađenih značajki na klasifikatorima istražena je i izmjerena kvaliteta pronađenih značajki.
Kako bi se značajke usporedile na realističnom skupu podataka, prikupljen je skup žanrovskih oznaka i preporuka
računalnih igara s platforme Steam. Prikupljen skup podataka zapisan je u obliku grafa.
Dobiveni rezultati višerazredne klasifikacije na značajkama različitih metoda analizirani su i uspoređeni korištenjem standardnih klasifikacijskih mjera.

Abstract (english)

In this thesis three modern methods for graph node feature encoding were analyzed; DeepWalk, LINE, Node2Vec.
Three different classifier types were used: logistic regression model, neareset centroid clustering and support vector machine.
Using the encoded node features to train the classifiers the quality of the different methods was explored and measured.
To compare the encoded features on a real-life dataset, a dataset of tags and suggestions of computer games on the Steam platform
was collected. The collected dataset was represented by a graph.
The results of the multi-label classification were analyzed and compared using common classification measures.

• Keywords: značajke grafova; duboko učenje; teorija grafova; teorija mreža; klasifikacija čvorova grafa; višerazredna klasifikacija
• Keywords (english): graph features; deep learning; graph theory; network theory; graph node classification; multi-label classification
• Language: croatian
• URN:NBN: urn:nbn:hr:168:845028
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-15 11:05:29