Učenje prijenosom za predikciju tipa ličnosti i demografskih informacija korisnika interneta

master's thesis

• Title (english): Transfer Learning for Personality and Demographics Prediction of Internet Users
• Author: Vinko Kašljević
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Marko Čupić (član povjerenstva)
• Committee member: Domagoj Jakobović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-07-10, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Predikcija tipa ličnosti i demografskih informacija je zanimljiv i koristan zadatak koji nam može pomoći u u boljem razumijevanju prirodnog jezika, kao i za razne društvene znanosti. U ovom radu smo isprobali višezadaćno učenje, specifično metode tvrdog i mekog dijeljenja parametara, na problemu predikcije tipova liˇcnosti i demografskih informacija. Također smo demonstrirali način učenja modela gdje treniramo na skupu podataka Reddit za koji nemamo sve oznake za svaki primjer iz skupa podataka. Rezultati pokazuju da ovakav pristup s danim značajkama i arhitekturom modela ne donosi značajna poboljšanja performansi modela u odnosu na učenje modela pojedinačno.

Abstract (english)

Predicting personality type and demographic information is an interesting and useful task that can help us in a better understanding of natural language as well as for various social sciences. In this paper, we tested multitask learning, specifically hard and soft parameter sharing metods, on the problem of personality and demographics prediction. Also, we demonstrated a way of learning a model where te train on a Reddit dataset for which we don’t have all the labels for each example in the dataset. The results show that this approach does not bring significant improvements in the performance of the model compared to learning the models individually.

• Keywords: duboko učenje; višezadaćno učenje; MBTI; Big Five; Enneagram; demografske informacije; tipovi ličnosti
• Keywords (english): deep learning; multi-task learning; MBTI; Big Five; Enneagram; demo- graphics; presonality types
• Language: croatian
• URN:NBN: urn:nbn:hr:168:894831
• 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 12:38:39