We are happy to invite you to start off the new academic year with the next seminar on “AI for Biomedicine” to be held on September 14th.
Speakers: Evgeni Levin (Horaizon) and Olivier J. M. Béquignon (AUMC).
Evgeni Levin (Horaizon): “MicrobiomeGPT“
Deep learning has revolutionized data science by effectively uncovering intricate structures in high-dimensional data across various domains. Despite its success, its application in certain areas, such as gut microbiome analysis, remains largely untapped. Transfer learning, a significant catalyst for deep learning’s success, improves performance on smaller datasets by harnessing knowledge from larger, analogous tasks. In this work, we propose a robust, pre-trained base model for fine-tuning on smaller datasets with related tasks. To this end, we’re created MicrobiomeGPT, a neural network pre-trained on an extensive corpus of microbial data. In comparison with tree-based models and directly trained neural networks, currently recognized as the state-of-the-art in microbial analyses, MicrobiomeGPT exhibits superior performance.
Olivier J. M. Béquignon (AUMC): “Development of a prediction model for blood-brain barrier penetrating drug-like molecules.”