Manifold Optimization in Data Analytics
Reshad Hosseini - Assistant Professor, University of Tehran
Sun, 9-May-2021 / 18:00 / Link:
Video Slides Poster


I will talk about manifold optimization and its application in data analytics. Two important manifolds covered in this talk are manifold of rotation matrices and manifold of symmetric positive definite matrices. I will talk about Quasi-newton and stochastic optimization methods on manifolds and present some results about them. Two important applications covered in this talk are statistical model fitting and pose graph optimization.


Reshad Hosseini received the B.Sc. degree in electrical engineering (telecommunication) from the University of Tehran, Tehran, Iran, in 2004, the M.Sc. degree in biomedical engineering (bioelectric) from the Amirkabir University of Technology, Tehran, in 2007, and the Ph.D. degree from the Faculty of Electrical Engineering and Computer Science, Technical University of Berlin, Berlin, Germany, in 2012. He did his Ph.D. research at the Max Planck Institute for Biological Cybernetics, Tuebingen, Germany. He is currently an Assistant Professor with the School of Electrical and Computer Engineering, College of Engineering, University of Tehran. His professional interest topics are machine learning, signal processing, and computational vision. He is particularly interested in the mathematical foundation of these fields, such as differential geometry, optimization, functional analysis, and statistics. His current research interests include manifold optimization, large-scale mixture models, 3-D reconstruction, neural system identification, visual recognition using deep learning, and accelerating reinforcement learning.