Klasifikacija čvorova u grafu metodama dubokog učenja

master's thesis

• Title (english): Node Classification in Graphs by Deep Learning Methods
• Author: Antonio Filipović
• Mentor: Tomislav Burić (mentor)
• Committee member: Tomislav Burić (predsjednik povjerenstva)
• Committee member: Josipa Pina Milišić (član povjerenstva)
• Committee member: Mate Puljiz (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Tema rada je primjena metoda dubokog učenja na klasifikaciju čvorova u grafu. Iz obrade signala matrice susjedstva proizašle su metode dubokog učenja koje imaju veliku primjenu u teoriji grafova i predstavljaju potencijalno veliki skok u uspješnosti rješavanja problema klasifikacije čvorova. Metode dubokog učenja specifično se treniraju samo za određeni zadatak, a prije njih postojale su metode koje koriste jedino strukturu grafa za sva daljna predviđanja. Cilj rada je implementirati i detaljno usporediti metode dubokog učenja (Graph Convolutional Networks, Graph Attention Networks i GraphSAGE) s metodom zasnovanom na strukturi grafa (Node2Vec) u rješavanju problema klasifikacije čvorova.

Abstract (english)

The theme of the master thesis is the application of deep learning methods to node classification in graphs. Most recent deep learning methods on graphs were born from the graph signal processing field and such methods present a potentially giant leap in the advancement of the graph machine learning field. Such methods utilize node features together with graph structure and are trained specifically for the task at the hand. Many deep learning methods before them only use graph structure and are trained for a broader range of problems. The goal of the master thesis is to implement and compare the most recent deep learning methods such as Graph Convolutional Networks, Graph Attention Networks, and GraphSAGE with older methods based only on graph structure such as Node2Vec.

• Keywords: klasifikacija čvorova; duboko učenje; grafovi; neuronske mreže
• Keywords (english): node classification; deep learning; graphs; neural networks
• Language: croatian
• URN:NBN: urn:nbn:hr:168:901283
• 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: 2023-01-13 14:53:22