Foundation model for HRMS data (MSc)

Level: MSc

Background

High resolution mass spectrometry (HRMS) can be used to identify unsafe food. At Wageningen Food Safety Research, we have gathered thousands of HRMS spectra over the years, which have been used to assess whether our food is safe. A better understanding of such data could lead to identification of unknown and emerging risks. However, labeling of unknown risks is impossible, and even if it were possible, unfeasible. Foundation models have shown enormous potential in grasping underlying concepts in text and images without needing specific labels. These concepts have also been used in medicine to identify patterns of disease (e.g. https://www.nature.com/articles/s42256-024-00807-9).

Research topic

In this project, you will investigate the potential of using foundation models for a specific type of HRMS data (HRMS-MS). You will review the theoretical necessities for applying such a model to data. Furthermore, you will implement a proof of concept foundation model on data of WFSR and / or public data. If successful, you will evaluate the results of such a foundation model to distinguish between safe and unsafe food.

What do we bring to the table?

You will be embedded in a group of ~ 20 data science and AI researchers, as well as HRMS domain experts. You have the opportunity to do your analyses on our high performance computing cluster.

What do we expect from you?

You have a background in computer science, artificial intelligence, or a related field. You have a strong programming background, and are proficient in python and pytorch.

Contact person(s)

Bedir Tekinerdogan (bedir.tekinerdogan@wur.nl)