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
-
02 July 2024
Start EngD programmes Autonomous Systems and Sustainable Process Design
In September, the Engineering Doctorate (EngD) programmes Autonomous Systems and Sustainable Process Design will start at the Faculty of Scfience and Engineering.
-
25 June 2024
Heineken Young Scientists Award for Casper van der Kooi
For his research in the field of Natural Sciences, Casper van der Kooi will receive the Heineken Young Scientists Award.
-
24 June 2024
Measuring stickiness
Several plant species use tiny sticky droplets to attract and trap insects. These droplets form an ideal toxin-free insecticide that could be easily washed off of edible plants. Abinaya Arunachalam built a tool to measure the stickiness of...