HIO; Representation Learning for Natural Language Processing (RL4NLP)
Kursusindhold
Representation Learning for Natural Language Processing (RL4NLP)
focuses on machine learning techniques for embedding language units
such as words, sentences, documents, and knowledge in a continuous
feature space. The course elaborates on methods for training such
features, applying them in downstream tasks, and interpreting them
from a computational linguistic point of view.
The course educates students on the most effective techniques, ranging from static word embedding methods to more advanced contextual embeddings and approaches based on attention mechanisms and language modeling. Following this technical grounding, students gain the theoretical insight and practical experience necessary to conduct research and develop a master’s thesis in this area.
The course is particularly relevant for students in the IT & Cognition program, computer science, and students from other disciplines with a sufficient mathematical and computer science background.
Representation Learning for Natural Language Processing (RL4NLP)
IT & Cognition
At the examination, the student can demonstrate:
Knowledge and understanding of:
- Theories and methods relevant to representing linguistic elements such as words, sentences, documents, and structured knowledge
- Problems related to embedding natural language elements into numerical feature spaces, including training and interpreting implicit patterns learned by deep learning models in NLP
Skills in:
- Discuss and document problems of relevance to language embeddings and representation learning
- Propose and evaluate solutions for relevant problems in representation learning
- Implement the foundational units of representation learning
Competencies in:
- Describing and analysing advanced topics within the field and applying them to real-world problems
- Designing, developing, and clearly documenting effective solutions to problems within the domain.
MA-level
2019 curriculum
See all the curriculums.
Lectures and class instructions, student presentations, and group/individual projects based on the number of students
- Linear Algebra (vector space, matrix operations, matrix
decomposition)
- Natural Language Processing (corresponding to the course Language
Processing)
- Deep learning (Pytorch, feed-forward neural and recurrent neural
networks)
- Programming experience in Python (corresponding to the course
scientific programming in IT & Cognition)
- ECTS
- 7,5 ECTS
- Prøveform
-
Skriftlig aflevering
- Prøveformsdetaljer
- Take-home assignment, optional subject
- Hjælpemidler
- Alle hjælpemidler tilladt
- Bedømmelsesform
- 7-trins skala
- Censurform
- Ingen ekstern censur
- Reeksamen
-
Conducted in the same manner as the original exam but can only be taken individually
- Kategori
- Timer
- Holdundervisning
- 28
- Forberedelse (anslået)
- 105
- Eksamensforberedelse
- 73
- Total
- 206
Kursusinformation
- Undervisningssprog
- Engelsk
- Kursusnummer
- HIOK0011EU
- ECTS
- 7,5 ECTS
- Niveau
- Master
- Varighed
-
1 semester
- Placering
- Efterår
- Skemagruppe
-
Uden for skemastruktur
- Studienævn
- Studienævnet for Nordiske Studier og Sprogvidenskab
Udbydende institut
- Institut for Nordiske Studier og Sprogvidenskab
Udbydende fakultet
- Det Humanistiske Fakultet
Kursusansvarlig
- Ali Basirat (4-6974716a48707d7536737d366c73)
Are you BA- or KA-student?
Courseinformation of students