Invited guest teacher

Dr. Matteo Fumagalli, University College London

Description

In this workshop I will firstly introduce basic concepts in machine learning and then move to explain how population genetic data can be presented to deep learning algorithms, namely neural networks. In the second part, in a hands-on practical, students will build a simple neural network to predict signals of natural selection. The overall intended learning outcome is to stimulate students to consider deep learning as part of their toolkit for population genetic inferences. 

Requirements

Students should be familiar with basic python and the use of jupyter notebooks. No prior knowledge of machine learning is needed. Knowledge of population genetics and evolutionary biology is desired. The guided exercise can be run on a laptop (after the installation of few packages, instructions will be given ahead of time) or on Google Colab (without any installation, but it requires a Google account). All the data needed will be made available.

The course will run on October 25th, 2024 from 10 am to 5 pm and will qualify for 0.8 hp.

To sign up for the course please contact the course leader Prof. Tanja Slotte (tanja.slotte@su.se) by email with the header “Deep learning population genetics”.