Beyond Clustering: Exploring and Interpreting Data with Topological Maps and the SOMTorch Library (TOPOMAP)
Organizers
- Andrés Ortiz García, University of Málaga. Email: andres.ortiz@uma.es
- Nicolás J. Gallego Molina, University of Málaga. Email: njgm@uma.es
- Carmen Jiménez Mesa, University of Málaga. Email: carmenj@uma.es
Description
During the tutorial, participants will progressively learn what a SOM (Self-Organizing Map) is, how its competitive learning works and why this method goes beyond a simple clustering algorithm. SOMs are a particularly useful tool for discovering patterns, revealing latent structures in data and providing interpretable representations of high-dimensional spaces.
Throughout the session we will build and train self-organizing maps step by step using the SOMTorch library. We will also explore the visualization utilities offered by the library to analyze the structure of the data and understand how it is organized on the map. In addition, we will show application examples in real problems such as data exploration, pattern detection, cluster analysis and interpretation of complex structures.
The goal is that attendees not only understand the theoretical foundations of Self-Organizing Maps, but also acquire the ability to apply them directly to their own data science and artificial intelligence problems.
The tutorial will have a hands-on format, in which participants will work with Python notebooks and real datasets to build, train and analyze SOM maps using SOMTorch.
