Modeli za objašnjenje rezultata predviđanja metoda strojnog učenja

master's thesis

• Title (english): Models for Explaining Predictions from Machine Learning Methods
• Author: Gregory Tomadin
• Mentor: Krešimir Pripužić (mentor)
• Committee member: Krešimir Pripužić (predsjednik povjerenstva)
• Committee member: Luka Humski (član povjerenstva)
• Committee member: Siniša Sovilj (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-14, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu opisana je objašnjiva umjetna inteligencije, kada i zašto se koristiti te čemu služi. Isto tako detaljno je opisana procedura stvaranja objašnjenja modela strojnog učenja od teorijske podloge pa sve do praktičnih primjera. U primjerima se koriste razni modeli strojnog učenja: linearna regresija, logistička regresija, stablo odluke, algoritam K najbližih susjeda, model slučajnih šuma itd. Osim toga koriste se razne metode objašnjive umjetne inteligencije: intrinzične metode, metode agnostičkih modela i metode specifične za model.

Abstract (english)

This paper describes explainable artificial intelligence, when and why to use it and what is its purpose. There is also a detailed description of the procedure for creating an explanation of the machine learning model from the theoretical background to practical examples. The examples use various machine learning models: linear regression, logistic regression, decision tree, K nearest neighbors, random forest, etc. and various explainable artificial intelligence metodes: intrinsic methods, model-agnostic methods and model-specific methods.

• Keywords: objašnjiva umjetna inteligencija; strojno učenje; objašnjivost; interpretabilnost; intrinzične metode; metode agnostičkih modela; metode specifične za model; metoda permutacijske važnosti; dijagram djelomične ovisnosti; SHAP
• Keywords (english): explainable artificial intelligence; XAI; machine learning; explainability; interpretability; intrinsic methods; model-agnostic methods; model-specific methods; permutation importance method; partial dependence plot; SHAP
• Language: croatian
• URN:NBN: urn:nbn:hr:168:967326
• 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 14:01:41