Active Learning for Improving Annotation Collection

Eligible: Masters students

Mentor: Michal Muszynski

Description: The mail goal of this project is to develop and extend existing statistical and active learning techniques to reduce generalization error of a machine learning model when candidate instances are selected at the training stage. Active learning intends to support data collection by automatically deciding which instances should be annotated to train a machine learning model in a quick and effective way.

Skills/Experience: Prior machine learning experience, intermediate skills in Matlab/Python/R programming, and willingness to learn new statistical and machine learning techniques required.

Contact: Interested students should send an email to Michal Muszynski with their CV and a short cover letter.