Statistical physics of learning vector quantization
PhD ceremony: Mr. A.W. Witoelar, 11.00 uur, Academiegebouw, Broerstraat 5, Groningen
Thesis: Statistical physics of learning vector quantization
Promotor(s): prof. M. Biehl, prof. N. Petkov
Faculty: Mathematics and Natural Sciences
The field of machine learning concerns the design of algorithms to learn and recognize complex patterns from data. Learning Vector Quantization (LVQ) constitutes an important family of such algorithms using prototypes which serve as typical examples. Despite its wide range of applications, the theoretical understanding of LVQ in general remains very limited. In this thesis a theoretical framework is constructed using concepts from statistical physics which allows for an exact analysis of the typical learning behaviors of the system. Studies in this thesis compare the characteristics of various LVQ algorithms to demonstrate the robustness of Neural Gas (NG) schemes and the specific advantages of LVQ 2.1 and Robust Soft LVQ algorithms. Furthermore surprising non-trivial behaviors are revealed including learning plateaus and phase transitions in the training process. The results provide insights to general prototype-based learning prescriptions.
Last modified: | 13 March 2020 01.13 a.m. |
More news
-
23 April 2024
Nine MSCA Doctoral Network grants for FSE researchers
Nine researchers of the Faculty of Science and Engineering have received a Horizon Europe Marie Sklodowska Curie Doctoral Network grant.
-
22 April 2024
Charissa Roossien secures JTF subsidy to develop Health Tracker
Dr. Charissa Roossien (ENTEG) has successfully secured a Just Transition Fund (JTF) subsidy of 1.8 million euros to develop a Health Tracker for reliable respiratory and metabolic analysis.
-
16 April 2024
UG signs Barcelona Declaration on Open Research Information
In a significant stride toward advancing responsible research assessment and open science, the University of Groningen has officially signed the Barcelona Declaration on Open Research Information.