HIOK00011U Specialization: Advanced Topics in Domain Adaptation

Volume 2015/2016
Content

Advanced Topics in Domain Adaptation

The course is an advanced course in Natural Language Processing and Machine Learning. Most work in this field assumes that training and test data points are randomly sampled from the same distribution. This assumption is almost always violated in practice, e.g., when learning general parsing models from annotated newswire, and model learning can be significantly improved by taking this into account. The course presents a wide range of methods for adapting models to new domains or distributions, incl., semi-supervised learning, importance weighting, representation learning, etc. The goal of the course is that all students implement a novel domain adaptation method with applications to one or more NLP tasks. The exam is a written report, taking the form of a conference-formatted research paper.

Holdundervisning om onsdagen fra kl. 09:30 - 12:00 på CST
  • Category
  • Hours
  • Class Instruction
  • 42
  • Course Preparation
  • 105
  • Exam Preparation
  • 59
  • Total
  • 206
Credit
7,5 ECTS
Type of assessment
Written examination
The exam is a written report, taking the form of a conference-formatted research paper.
Censorship form
External censorship