DDLS course > Course information
Hide menu

DDLS course Transfer Learning for Life Science

Course information

Transfer learning and life science

Transfer learning is a branch of machine learning that enables transferring information from one knowledge domain into another related domain. This methodology is particularly valuable in the data driven life science, for example in medical imaging and transcriptomics, where labelled data can be scarce and/or expensive to obtain. At the same time, there are many publicly available repositories containing valuable information that can be transferred into related biomedical contexts. Transfer learning makes it possible to transfer information across cell populations, instruments, medical institutions and different data types and allows for greatly reducing the resources needed to develop accurate predictive models. The course will explore different transfer learning paradigms and models at a conceptual level, will introduce the relevant software tools, and will study use cases from domains such as medical imaging, omics, and clinical data.

Course description

Course covers 5 ECTS credits and is given for PhD students from DDLS Research School. The course includes lectures, practical sessions (labs), project work, project presentation and opposition. One full week is given on campus, rest is online.

The syllabus of the course can be found here

Literature

Course overview

Topic 1a: Overview of architectures for transfer learning: feed-forward networks, convolutional networks, adversarial networks, autoencoders, transformers, contrastive deep learning, self-supervised learning.

Reading materials

Lecture slides

Topic 1b: Introduction to transfer learning and challenges in life science data. Taxonomy of transfer learning. Software for transfer learning.

Reading materials

Lecture slides

Practical session materials


Page responsible: Oleg Sysoev
Last updated: 2026-08-26