ML for Scientific Data: From Classical ML to Deep Learning¶
Online training workshop: ML for Scientific Data: From Classical ML to Deep Learning, 8 - 9 December, 2026
Learn how to
- explore and visualise real-world scientific datasets in Python
- extract features from raw data
- train and evaluate a Random Forest classifier
- build, train and evaluate a CNN for image segmentation
- submit ML training jobs to an HPC cluster with Slurm.
The examples use data from published research: eye-tracking recordings and microscopy images of titanium alloy. The course is at the beginners/intermediate level and consists of lectures interspersed with hands-on sessions. The training is organised by NAISS, with contributions from experts from Sweden AI Factory. The hands-on exercises will use NAISS's Arrhenius system.
Prerequisites: basic Python (numpy/pandas), the Linux command line and submitting a job on cluster. A course project is provided for those without HPC access.
Dates and location¶
- Day 1 (8 December 2026), 10-15, Introduction to Random Forest: eye-tracking data, user identification and gender classification
- Day 2 (9 December 2026), 10-15, Introduction to Convolutional Neural Networks: microscopy images, pore segmentation
The workshop will be held via zoom. A zoom link will be provided to registered participants prior to the event.
Links¶
More information and material for the ML for Scientific Data workshop
Source repository