Robust Multidimensional Scaling Using a Maximum Correntropy Criterion -- talk by Fotios Mandanas (Thessaloniki) Event
- Time:
- 14:00 - 16:00
- Date:
- 15 November 2018
- Venue:
- Lecture Theatre 8C, Room 8031, Building 54, Mathematical Sciences, University of Southampton, Highfield Campus, SO17 1BJ
For more information regarding this event, please email Dr Konstantinos Katsikopoulos at K.Katsikopoulos@southampton.ac.uk .
Event details
Multidimensional scaling (MDS), which refers to a class of dimensionality reduction techniques, has been exploited in order to visualize in a reduced dimensional metric space the hidden structures among a set of entities, so that the interpoint distances in this space approximate the initial pairwise dissimilarities between entities as closely as possible. Here, we are interested in cases whenever the initial dissimilarity matrix is contaminated by outliers. The traditional methods for solving MDS are vulnerable to outliers inducing highly corrupted embeddings. To remedy this vulnerability, a unified framework for the solution of MDS problem is proposed, where the MDS is treated as maximization of a correntropy criterion, which is solved by half-quadratic optimization in either multiplicative or additive forms. The aforementioned framework employs potential functions of M-estimators in combination with Frobenius or l2,1-norm or nuclear norm regularization. Novel algorithms are derived for each case. Their performances are assessed experimentally against three state-of-the-art MDS techniques on various benchmark data sets under the same conditions. The experimental results demonstrate that the proposed algorithms perform better than the competing alternatives.
Speaker information
Fotios Mandanas,350 Guided Missile Wing, Thessaloniki, Greece,Fotios D. Mandanas was born in Serres, Greece, in 1976. He received the Bachelor of Science Degree in telecommunications and electronics engineering from the Hellenic Air Force Academy, Dekelia, Greece in 1999. He received the first Master of Science degree in State-of-the-Art Design and Analysis Methods in Industry (track Production Management and Industrial Administration) from the Department of Mechanical Engineering of the University of Thessaly, Volos, Greece in 2005 and the second Master of Science degree in Informatics and Communications with specialization to Digital Media and Computational Intelligence from the Department of Informatics of Aristotle University of Thessaloniki, Thessaloniki, Greece, where since December 2014, he has been working toward the Ph.D. degree with the same department. He worked in 111 Combat Wing, Nea Anchialos, Greece from 1999 to 2011 where he participated in operational, intermediate and depot level maintenance of avionics systems. In 2001, he participated in a training in Boston/Massachusetts and Baltimore/Maryland regarding the hardware and software development of interface test adapters for the control of printed circuit boards. He worked in 2012 in Weapon Systems Department of the Air Force Support Command, Elefsis, Greece where he participated in the support and maintenance of Weapon Systems and the implementation of the respective contracts. Then, he was in Research and Development Directorate of Hellenic Air Force Electronics Depot, Glyfada, Athens. He is currently Chief engineer in 350 Guided Missile Wing, Thessaloniki, Greece. His research interests include signal processing, machine learning, computational intelligence, and graph theory. He was a Scholar of the State Scholarships Foundation of Greece and the Department of Informatics of Aristotle University of Thessaloniki.