Plenary Lecture




Prof. Theodore B. Trafalis
School of Industrial and Systems Engineering, The University of Oklahoma, Oklahoma, USA


Title: Federated Learning: Algorithmic Challenges under Privacy Restrictions and Weather Applications with Imbalanced Data

Abstract: Federated learning is a distributed machine learning paradigm that allows for the collaborative training of models across various agents without directly sharing sensitive data, ensuring privacy and robustness. In this talk I discuss federated learning, investigating the algorithmic and data-driven challenges of deep learning models in the presence of additive noise in this framework. In addition strategies to measure the generalization, stability, and privacy-preserving capabilities of these models are discussed. Specifically, five noise infusion mechanisms at varying noise levels within centralized and federated learning settings are explored. As model complexity is a key component of the generalization and stability of deep learning models during training and evaluation, a comparative analysis of three Convolutional Neural Network (CNN) architectures is provided. Several metrics for training with noise are also discussed. Signal-to-Noise Ratio (SNR) is introduced as a quantitative measure of the trade-off between privacy and training accuracy of noise-infused models, aiming to find the noise level that yields optimal privacy and accuracy. The Price of Stability and Price of Anarchy are also defined in the context of privacy-preserving deep learning, contributing to the systematic investigation of the noise infusion mechanisms to enhance privacy without compromising performance. I also discuss a real-world application of federated learning in weather prediction applications that suffer from the issue of imbalanced datasets. Using data from multiple sources combined with advanced data augmentation techniques show that the accuracy and generalization of weather prediction models is improved.

Bio: Theodore B. Trafalis, PhD, is a Professor at the School of Industrial and Systems Engineering , adjunct professor in the School of meteorology and the Data Science Institute at the University of Oklahoma, USA. He earned his BS in mathematics from the University of Athens, Greece, his MS in Applied Mathematics, MSIE, and PhD in Operations Research from Purdue University. He is a member of INFORMS, SIAM, International Society of Multiple Criteria Decision Making, and the International Society of Neural Networks. He has been listed in several Who’s Who biographies such as in the 1993/1994 edition of Who’s Who in the World. He was a visiting Assistant Professor at Purdue University (1989-1990), an invited Research Fellow at Delft University of Technology, Netherlands (1996), a visiting Associate Professor at Blaise Pascal University, France, and at the Technical University of Crete (1998). He was also an invited visiting Associate Professor at Akita Prefectural University, Japan (2001). The academic year 2006-2007 was on a sabbatical at the National Center for Scientific Research “Demokritos”, Institute of Informatics and Telecommunications, Computational Intelligence Laboratory (CIL), Athens, Greece. In June 2011 was invited researcher at the Institute of Applied Mathematics, University of Toulouse, France. In March 2014 he was invited professor at the Department of Computing, Unitec, Auckland, New Zealand. In the academic year was on a sabbatical visiting the school of Electrical and Computer Engineering, National Technical University of Athens, Greece. From 2014 to 2018 participated in joint research with LATNA, Higher School of Economics, Nizhny Novgorod, Russia.His research interests include: operations research/management science, mathematical programming, interior point methods, multiobjective optimization, control theory, artificial neural networks, kernel methods, evolutionary programming data mining, global optimization , Machine Learning, AI and weather applications. He has published more than one hundred fifty articles in journals, conference proceedings, edited books, made over one hundred fifty technical presentations, and received several awards for his papers. In 2004 he received the Regents Award at the University of Oklahoma for his research activities. He has been funded through National Science Foundation (NSF), NOAA, Office of the Army and received the NSF Research Initiation Award in 1991. In 2006 he was the editor of a special issue in Support Vector Machines for the journal of Computational Management Science. He also co-edited a special issue in “Learning from Data” for the same journal. He has coauthored a research monograph with Springer under the title “Robust Data Mining” that was published in 2013. Prof. Trafalis currently serves as the chief editor of Intelligent Control and Automation, an associate editor for the Journal of Heuristics and several other journals. In addition he has been on the Program Committee of several international conferences in the field of intelligent systems, AI, data mining and optimization. He was also co-organizer of the International Conference on the Dynamics of Disasters, Athens, Greece, 2006 and organizer of the International Conference in Industrial Systems and Design Engineering that is held in Athens, Greece every June since 2011.


Contact

Venue

The Golden Age Hotel of Athens
57 Michalakopoulou Street, 115 28 Athens Greece
+302107240861
contact@hotelgoldenage.com
https://hotelgoldenage.com/

Email Us

info@mmctse.org, mmctse.conference@gmail.com

Call Us