This is the course page for the seminar course Evaluation of NLP systems: metrics, methods, and practices at the Department of Linguistics, University of Tübingen.
Correct evaluation of machine learning systems is important for both verifying new models/methods and for choosing a system to use for a particular purpose. In this course, we will study a broad set of issues with evaluation, such as what metrics to use; how to compare alternative models correctly, confidently; how to make sure our results and conclusions are verifiable.
This course is a typical seminar course. After introductory lectures during the first few weeks, participants are required to read, discuss and present selected papers from literature. We will also do hands-on demonstrations of some of the concepts through a “shared task”.
The normal course work is worth 6ECTS. It is possible to obtain 9ECTS with additional work (typically a term paper).
Requirements
Earlier machine learning exposure is essential for success in this course. Completion of ‘Statistical Language Processing’ or similar course work or experience is required.
Contact
- Instructor: Çağrı Çöltekin
<ccoltekin@sfs.uni-tuebingen.de>
Office hours: Monday 16:00 - 17:00