Otkrivanje anomalija u velikim skupovima podataka

master's thesis

• Title (english): Anomaly Detection in Big Datasets
• Author: Bruno Rosan
• Mentor: Mirko Randić (mentor)
• Committee member: Mirko Randić (predsjednik povjerenstva)
• Committee member: Bruno Blašković (član povjerenstva)
• Committee member: Mihaela Vranić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2017-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U sklopu rada sustavno su opisane vrste anomalija te poznate metode njihovog otkrivanja, uz naglasak na metode strojnog učenja. Za sam početak dan je kratak uvod u područje detekcije anomalija i njezine uloge u analizi velikih skupova podataka. U prvom poglavlju upoznaje se s osnovnim pojmovima vezanim za problem otkrivanja anomalija, kao i različitim pristupima riješavanja problema. Slijedi detaljan opis metoda korištenih u radu koje se uglavnom temelje na tehnikama strojnog učenja. U nastavku su opisana dva javno dostupna skupa podataka korištena u sklopu rada. Četvrto poglavlje daje pregled provedenih eksperimenata zajedno s dobivenim rezultatima koristeći standardne mjere uspješnosti (ako se radilo o označenim podacima). Također objašnjava se motivacija koja stoji iza korištenja određenog pristupa za konkretan skup podataka. Za kraj se iznosi kratak osvrt i zaključak na dobivene rezultate.

Abstract (english)

Throughout the thesis, the types of anomalies and known methods of their detection are systematically described, with an emphasis on machine learning methods. At the begining a short introduction to the area of anomaly detection and its role in the data mining is given. The first chapter introduces the basic concepts related to the anomaly detection problem, as well as various approaches to solving this problem. After that, a detailed description of the methods used in work is given, mainly based on machine learning techniques. Description of two publicly available data sets used in the thesis is given next. The fourth chapter gives an overview of the experiments carried out together with the results obtained using standard performance measures (in case of labeled data). Motivation for using specific approach to a concrete dataset is also explained. Last chapter gives a brief review and a conclusion over obtained results.

• Keywords: Anomalija; novitet; detekcija; klasifikacija
• Keywords (english): Anomaly; novelty; detection; classification
• Language: croatian
• URN:NBN: urn:nbn:hr:168:346929
• 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: 2019-03-14 00:30:53