AI/ML knowledge base

The usefulness of artificial intelligence/machine learning technologies in neuroscience spans from the practical to the abstract; they can be used to automatize labor-intensive analysis steps or for explorative analysis of large, complex data sets. They can even serve as model systems to study emergent properties of neural networks.

The knowledge base can help you to

• get educated in AI/ML applications through MP-internal workshops
• get free access to large-scale computing

Project example: image segmentation

Artificial neural networks were trained to segment out positively labelled cells in images. The model, source code, and a user tutorial are available on the Microsoft Teams page (“AI/ML knowledge base (MultiPark)”) and on GitHub: https://github.com/NRC-Lund/multipark-aiml. 

To better coordinate AI/ML efforts within MultiPark, we are implementing the data management principles of the Open Metadata Initiative for Neuroscience Data Structures (openMINDS) for data generated by selected MultiPark research groups and infrastructures. This will make existing datasets findable, accessible, interoperable and reusable for improved training of AI models. OpenMIND is managed by EBRAINS, a larger effort to develop shared digital neuroscience infrastructures and services funded by the EU. MultiPark is currently working towards joining the Swedish node of EBRAINS. 

Please get in touch with par [dot] halje [at] med [dot] lu [dot] se (Pär Halje) if you think this sounds interesting.