Graph generative models are recently gaining significant interest in current application domains. They are commonly used to model social networks, knowledge graphs, molecules and proteins. In this talk we will present the potential of graph generative models and our recent relevant efforts in the biomedical domain. More specifically we present a novel architecture that generates medical records as graphs with privacy guarantees. We capitalize and modify the graph Variational autoencoders (VAEs) architecture. We train the generative model with the well known MIMIC medical database and achieve generated data that are very similar to the real ones yet provide privacy guarantees. We achieve there as well promising results with potential for future application in broader biomedical related tasks. Finally we present ongoing research directions for multi modal generative models involving graphs and applications to protein function text generation – the prot2text model.

 

 

Michalis Vazirgiannis is a Distinguished Professor at Ecole Polytechnique in France. He has been intensively involved in data science and AI related research. His broad research area is in methods for data mining and machine/deep learning methods for diverse data types and applications (including graphs, text, time series). Recently he is working of i. GNNs and aspects including expressiveness, efficiency, generation ii. Pretrained models and resources for multilingual NLP and Biomedical applications. His research and industrial impact is spanning different domains such as web advertising, social networks, online gambling, insurance, legal text applications, aviation and maritime industry and the bio/medical domain. He has been teaching machine/deep learning and AI courses for academic and executive training studies in France, China, UAE, Morroco, Spain, Greece. Pr. Vazirgiannis has published more than 250 papers in international journals and proceedings of international conferences including NeurIPS, AAAI/IJCAI, AISTATS, ICLR, EMNLP/ACL, JMLR, JAIR, KDD, Nature/Digital Medicine, Scientific Reports. On the side of supervision he has supervised 30 completed PhD theses. Finally he has been able to attract significant funding for research from national and international sources, from research agencies and industrial partners (including Google, Airbus, Huawei, Deezer, BNP, LVMH). He lead(s) academic research chairs (DIGITEO 2013-15, ANR/HELAS 2020-26) and an industrial one (AXA, 2015-2018). He has received several awards and distinctions including i. Marie Curie Intra European Fellowship (2006-8) ii. “Rhino-Bird International Academic Expert Award” in recognition of his academic/professional work @ Tencent (2017), iii. best paper awards in international conferences (such as IJCAI 2018, CIKM2013). He has been invited to media interviews in France, USA and China and published popularized articles in French and Greek magazines/newspapers on Artificial Intelligence topics.
More info at: http://www.lix.polytechnique.fr/dascim