Computational Simulations of Language

Dit is een conceptversie. De vakomschrijving kan nog wijzigen, bekijk deze pagina op een later moment nog eens.

Faculteit Science and Engineering
Jaar 2022/23
Vakcode WMCC009-05
Vaknaam Computational Simulations of Language
Niveau(s) master
Voertaal Engels
Periode semester II a
ECTS 5
Rooster rooster.rug.nl

Uitgebreide vaknaam Computational Simulations of Language
Leerdoelen At the end of the course, students should be able to talk about the history of computational simulations of language learning, and understand the differences between learning models, recognizing the benefits as well as the shortcoming. Students should also be able to set up computational simulation of a statistical learning experiment when presented with experimental data.
Omschrijving This course introduces students to implicit statistical learning in language, and computational simulations of that learning. The course also gives an introduction to the debate between empirical and rational models of language.

Topics covered include:

1. Connectionist models of language: their rationale, achievements and drawbacks
2. The debate between connectionism and symbolic models of cognition
3. Psychological proposals of learning applied to language, including naive discriminate learning
4. The similarities and differences in learning progression modelled with recurrent neural networks compared to modelling with naive discriminate learning.
5. Insight into what statistical language patterns can be learned implicitly by infants and adults, what statistical patterns are challenging, and what this tells us about human cognition

Student evaluations are based on the results of three computer labs. In these labs students:

• recreate and investigate the behavior of classic connectionist models.
• recreate and investigate the behavior of naïve discriminate learning
• model experimental data that illustrates statistical learning in humans, comparing naïve discriminate learning and compare these models to neural network models
• evaluate the results of the simulations compared to results from human participants

The course gives a gentle introduction into simple neural networks and naïve discriminant learning. Students are not expected to have previous experience with neural networks or other machine learning methods in order to do the course.
Uren per week
Onderwijsvorm Hoorcollege (LC), Practisch werk (PRC)
(Introductory lecture, homework and lab sessions)
Toetsvorm Opdracht (AST)
Vaksoort master
Coördinator Dr. J.K. Spenader
Docent(en) Dr. J.K. Spenader
Entreevoorwaarden Mandatory: No prior knowledge is assumed. Please note that the student is expected to have a relevant BSc degree.
Opmerkingen
Opgenomen in
Opleiding Jaar Periode Type
MSc Artificial Intelligence  (C - Elective Course Units) - semester II a keuze
MSc Computational Cognitive Science  (B - Mandatory Course Units Cognitive Language Modelling) 1 semester II a verplicht CLM
MSc Computational Cognitive Science  (C - Elective Course Units) - semester II a keuze
MSc Courses for Exchange Students: AI - Computing Science - Mathematics - semester II a
ReMa Taalwetenschappen / Linguistics  (ReMa Language and Communication Technologies (LCT); Erasmus Mundus) 1 semester II a keuze