RIA – Institutul de Cercetare în Inteligență Artificială, Realitate Virtuală și Robotică anunta expunerea dnei Prof. Dr. Mandy Lange-Geisler (Hochschule Mittweida) cu titlul:

An Introduction to Hebbian Learning Approaches for Prototype-based Classification and Generalized Learning Vector Quantization with Selected Applications

Expunerea va avea loc miercuri, 15.1.2025, sala C335, incepand cu ora 12.30.

Abstract:
Hebbian learning approaches are the basis for many machine learning algorithms.  For instance, Hebb’s postulate of learning is incorporated into Learning Vector Quantization (LVQ) algorithms, which belong to the prototype-based classification methods. In LVQ, the adaptation of so-called prototypes is realized by the Hebbian principle. LVQ algorithms are among the most successful classifiers with numerous variants and extensions, such as Generalized Matrix Learning Vector Quantization (GMLVQ). These extensions incorporate advanced learning schemes, including adaptive metric learning, which not only enhances classification accuracy but also provides valuable interpretability of the model. This is useful for many practical classification problems.