Korištenje naprednih tehnika dohvatom-pojačanog generiranja za prepoznavanje futuroloških signala

undergraduate thesis

• Title (english): Utilization of advanced retrieval-augmented generation techniques for futures signal recognition
• Author: Igor Šoštarko
• Mentor: Mario Brčić (mentor)
• Committee member: Mario Brčić (predsjednik povjerenstva)
• Committee member: Mihael Kovač (član povjerenstva)
• Committee member: Kristijan Poje (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2024-07-05, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Kod korištenja velikih jezičnih modela je značajan problem ustajalo znanje modela, što može dovesti do neoptimalnih rezultata, posebice kod problema koji zahtijevaju pračenje najnovijih informacija, poput prepoznavanja futuroloških signala. U ovom radu je predloženo riješenje tog problema korištenjem dohvatom-pojačanog generiranja za čiju se implementaciju koristi vektorska baza podataka. Kao način evaluacije uspješnosti takvog pristupa se provodi usporedba performansa takve implementacije naspram performanse samog jezičnog modela na zadatku prepoznavanja futuroloških signala. Takav je pristup pokazao obečavajuće rezultate, dajući bolje rezultate od samog jezičnog modela, no i određene nedostatke kao što je smanjenje točnosti klasificiranja članaka koji su signali.

Abstract (english)

One significant issue with using large language models is the problem of stale knowledge, which can lead to sub-optimal results, especially for tasks that require tracking the latest information, such as recognizing futures signals. This paper proposes a solution to this problem by using Retrieval-Augmented Generation implemented with a vector database. To evaluate the success of this approach, the performance of this implementation is compared to the performance of the language model alone in the task of recognizing futures signals. Such an approach has shown promising results, yielding better outcomes than the language model alone, but also certain shortcomings, such as a decrease in the accuracy of classifying articles that are signals.

• Keywords: dohvatom-pojačano generiranje; jezični model; umjetna inteligencija; futurološki signal
• Keywords (english): Retrieval-Augmented Generation; Language Model; Artificial Intelligence; Futures Signal
• Language: croatian
• URN:NBN: urn:nbn:hr:168:728665
• 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: 2024-10-12 22:17:56