Skip to content

Latest commit

 

History

History
74 lines (48 loc) · 2.2 KB

README.md

File metadata and controls

74 lines (48 loc) · 2.2 KB

Automating Science

The repository for the course "Automating Science" taught at the Jagiellonian University in 2024. The page is hosted by the machine learning research group GMUM.

Theme

How Machine Learning can facilitate scientific discoveries. We will discuss how we make progress and potential role of applying AI methods into the scientific problems.

Schedule

  • lectures: Wednesdays 4pm
  • tutorials: Fridays 10.15am

We highly encourage you to attend the classes and engage with your colleagues.

Lecture content

Grading

Exam 25% at the end of semester

Project 50% composed of:

  • first stage 20 %
  • mid stage 20 %
  • final stage 60 %

Homework 15% where:

  • top 2 notebooks 50%
  • rest notebooks 50%

Piazza / lecture / tutorial activity 10%

Piazza

Piazza Automating Science 2024

Contact

If possible, by Piazza. If not successful, then reach us by emails.

Set up

data directory (added to gitignore) will store the data: mnist, fashionmnist, cifar, synbols models directory (added to gitignore) will store the pretrained / results of trained models jtt directory for just-train-twice (week 2) all_about_uncertainty (weeks 3,4)

Install conda (follow the official instructions )

conda env create -f environment.yml

Info

(please do not complete the notebooks before the tutorials)

  • Week 1: introduction
  • Week 2: shortcut learning
  • Week 3: uncertainty
  • Week 4: uncertainty and anomaly detection
  • Week 5: [remote] Project discussions
  • Week 6: transformers
  • Week 7: continual learning
  • Week 8: drug discovery
  • Week 9+: into the wild / project meetings