Algoritmi za brzo učenje na neprijateljskim primjerima

undergraduate thesis

• Title (english): Algorithms for fast adversarial learning
• Author: Dominik Jambrović
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Ivan Grubišić (član povjerenstva)
• Committee member: David Dukić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-06, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Učenje s neprijateljskim primjerima važno je za postizanje sigurnih modela. Proučavamo osnovne načine učenja s neprijateljskim primjerima (FGSM, PGD). Uočavamo i istražujemo probleme prisutne pri korištenju PGD načina učenja na kompleksnijim modelima. Razmatramo predložene nadogradnje na osnovne načine učenja s neprijateljskim primjerima (FreeAdv, FastAdv, FastAdv+, FastAdvW) i evaluiramo performanse naučenih modela. Evaluacija pokazuje da najbolje performanse postižu modeli učeni FastAdv+ i FastAdvW načinima učenja.
Razmatramo mogućnost korištenja robusnih modela za detekciju zatrovanih podataka.

Abstract (english)

Adversarial learning is important for achieving secure models. We study the basic adversarial learning methods (FGSM, PGD). We observe and investigate the problems present when using the PGD learning method on more complex models. We consider proposed upgrades to the basic adversarial learning method (FreeAdv, FastAdv, FastAdv+, FastAdvW) and evaluate the performance of the trained models. The evaluation shows that the best performance is achieved by models trained using the FastAdv+ and FastAdvW learning methods.
We consider the possibility of using robust models for the detection of poisoned data.

• Keywords: robusni modeli; učenje s neprijateljskim primjerima; FGSM; PGD; besplatno učenje s neprijateljskim primjerima; brzo učenje s neprijateljskim primjerima; zatrovani podatci
• Keywords (english): robust models; adversarial learning; FGSM; PGD; free adversarial learning; fast adversarial learning; poisoned data
• Language: croatian
• URN:NBN: urn:nbn:hr:168:753638
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-07-13 10:08:24