Javascript must be enabled for the correct page display

The Role of Context in Object-Word Learning: Findings from Computational Modeling

van Buijtene, Tinke (2019) The Role of Context in Object-Word Learning: Findings from Computational Modeling. Bachelor's Thesis, Artificial Intelligence.

[img]
Preview
Text
AI_BA_2019_TinkevanBuijtene.pdf

Download (2MB) | Preview
[img] Text
toestemming.pdf
Restricted to Registered users only

Download (120kB)

Abstract

It is yet unknown how exactly children are able to match objects they perceive with their correct labels. This is known as the mapping problem. Researchers have proposed various theories on how this problem could possibly be dealt with, two of the most well-known being cross-situational learning and propose-but-verify. But one aspect is often overlooked: the context in learning environments. For this study, I have built computational models of cross-situational learning and propose-but-verify that take on experiments done by Dautriche and Chemla (2014). Using these models gave me the opportunity to inspect closely how different contextual setups in the experiments affect the performance of both strategies. The models’ performances were then compared to human data. Results showed that even with the context modulations, neither necessarily resembles human data beUer than the other. I discuss one possible explanation for this finding: that people actually use a combination of both strategies.

Item Type: Thesis (Bachelor's Thesis)
Supervisor name: Spenader, J.K. and Rij-Tange, J.C. van
Degree programme: Artificial Intelligence
Thesis type: Bachelor's Thesis
Language: English
Date Deposited: 28 Feb 2019
Last Modified: 04 Mar 2019 10:09
URI: https://fse.studenttheses.ub.rug.nl/id/eprint/19219

Actions (login required)

View Item View Item