Prof. I. Tsamardinos’ lecture at Hikone Data Science on Causal Discovery
Dr. Ioannis Tsamardinos, CEO at JADBio, and Professor at the Computer Science Department at the University of Crete, Greece, spoke among other significant speakers, at the International Symposium on Causal Discovery & Machine Learning, during the Hikone Data Science 2021 conference, September 10-11, 2021. The Symposium covered a wide range of topics, with Dr. Tsamardinos lecturing about his work on causality, including causal discovery with mixed data, biomedical applications, and software CAUSAL PATH.
Find the title and the abstract of his presentation, and watch the video recording below.
Title: “Towards Automated Causal Discovery and Learning Causal Dynamical Systems”
Abstract: There is no causal discovery algorithm that is superior to all other causal discovery problem algorithms. The optimal algorithm to use depends on the size and type of data at hand. However, unlike standard predictive modeling, causal discovery is unsupervised. Hence, choosing the optimal algorithm cannot trivially use techniques like cross-validation. In the first part of the talk, we’ll present the OCT methodology, standing for Out-of-Sample Causal Tuning, that heuristically selects the best combination of causal discovery algorithm and its hyper-parameter values. OCT and similar algorithms could lead to Automated Causal Discovery Systems that fully automatically perform a causal analysis of a dataset. In the second part of the talk, we’ll focus on learning causal models in the form of dynamic systems, such as differential equation systems. We’ll present the Unified Sparse Dynamics Learning (USDL) algorithm that can learn both discrete and continuous-time systems. USDL introduces the weak formulation technique that removes the need to approximate derivatives and could be useful to many similar algorithms.
Speaker Bio:
Dr. Ioannis Tsamardinos acquired his Ph.D. in 2001 from Pittsburgh University, USA. He worked as an Assistant Professor at the Department of Biomedical Informatics, Vanderbilt University, until 2006 when he joined the University of Crete. His research interests are Artificial Intelligence and Philosophy of AI, Artificial Intelligence in Biomedicine, Machine Learning, Causal Inference, and Induction, Learning from Biomedical Data, Feature and Variable Selection for Classification, Bioinformatics, Planning, Applications of Machine Learning in Biomedical Informatics.
Start using autoML with a FREE account at https://jadbio.com
