prikaz prve stranice dokumenta Detekcija anomalija u vremenskim serijama
No public access
master's thesis
Detekcija anomalija u vremenskim serijama
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2024. urn:nbn:hr:168:964023
Poletan, Nino
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Poletan, N. (2024). Detekcija anomalija u vremenskim serijama (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:964023

Poletan, Nino. "Detekcija anomalija u vremenskim serijama." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2024. https://urn.nsk.hr/urn:nbn:hr:168:964023

Poletan, Nino. "Detekcija anomalija u vremenskim serijama." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2024. https://urn.nsk.hr/urn:nbn:hr:168:964023

Poletan, N. (2024). 'Detekcija anomalija u vremenskim serijama', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 13 January 2025, https://urn.nsk.hr/urn:nbn:hr:168:964023

Poletan N. Detekcija anomalija u vremenskim serijama [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2024 [cited 2025 January 13] Available at: https://urn.nsk.hr/urn:nbn:hr:168:964023

N. Poletan, "Detekcija anomalija u vremenskim serijama", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2024. Available at: https://urn.nsk.hr/urn:nbn:hr:168:964023

Metadata
TitleDetekcija anomalija u vremenskim serijama
Title (english)Anomaly Detection in Time Series
AuthorNino Poletan
MentorVladimir Čeperić (mentor)
Committee memberVladimir Čeperić (predsjednik povjerenstva)
Committee memberStjepan Groš (član povjerenstva)
Committee memberTomislav Marković (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2024-09-20, Croatia
Abstract
Ovaj rad istražuje detekciju anomalija u vremenskim serijama koristeći različite algoritme i modele, s posebnim naglaskom na velike jezične modele (LLM) poput GPT-4o i GPT-4o-mini. Detekcija anomalija je ključna u raznim industrijama jer omogućuje prepoznavanje neočekivanih obrazaca koji mogu ukazivati na pogreške, prijevare ili sigurnosne prijetnje. Rad analizira tradicionalne metode detekcije anomalija kao što su Isolation Forest, LOF, AutoEncoder, i VAE, te nove pristupe koji koriste LLM-ove. Rezultati pokazuju da iako LLM-ovi imaju potencijal za detekciju anomalija, njihova učinkovitost je ograničena bez specifične prilagodbe i treninga na relevantnim skupovima podataka. Korištenje zero-shot pristupa za detekciju anomalija pokazalo se najmanje učinkovitim, dok su druge metode poput biblioteke LLMTime pokazale potencijal za daljnji razvoj.
Abstract (english)
This thesis explores anomaly detection in time series data using various algorithms and models, with a specific focus on large language models (LLMs) such as GPT-4. Anomaly detection is crucial in various industries as it allows for the identification of unexpected patterns that may indicate errors, fraud, or security threats. The study analyzes traditional anomaly detection methods such as Isolation Forest, LOF, AutoEncoder, and VAE, as well as new approaches utilizing LLMs. The results indicate that while LLMs have potential for anomaly detection, their effectiveness is limited without specific tuning and training on relevant datasets. The use of zero-shot approaches for anomaly detection proved to be the least effective, whereas other methods, like the LLMTime library, demonstrated potential for further development.
Languagecroatian
URN:NBNurn:nbn:hr:168:964023
Study programmeTitle: Computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2024-10-12 21:53:23