1 / 10

Self-Organizing Map (SOM) = Kohonen Map

Self-Organizing Map (SOM) = Kohonen Map. artificial neural network topological order 2D lattice of neurons. Self-Organizing Map (SOM) = Kohonen Map. artificial neural network topological order 2D lattice of neurons training with n -dimensional data

piper
Télécharger la présentation

Self-Organizing Map (SOM) = Kohonen Map

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Self-Organizing Map (SOM) = Kohonen Map • artificial neural network • topological order • 2D lattice of neurons

  2. Self-Organizing Map (SOM) = Kohonen Map • artificial neural network • topological order • 2D lattice of neurons • training with n-dimensional data • e.g., census statistics; financial data; text documents

  3. Self-Organizing Map (SOM) = Kohonen Map • artificial neural network • topological order • 2D lattice of neurons • training with n-dimensional data • e.g., census statistics; financial data; text documents

  4. Self-Organizing Map (SOM) = Kohonen Map

  5. Self-Organizing Map (SOM) = Kohonen Map • SOM as dimensionality reduction method

  6. Self-Organizing Map (SOM) = Kohonen Map • SOM as clustering method

  7. Self-Organizing Map (SOM) = Kohonen Map

  8. SOM – Examples of Edge Effects • Higher density representation of n-dimensional space at edge neurons • Indicated by: • Higher density of input vectors mapped onto edge neurons • Higher internal variance of input vector mapped onto edge neurons • Criss-crossing of trajectories from one edge to another • what I call the Siberia-Alaska effect in world maps centered on prime meridian (they’re close in geographic space, yet widely separated in world map) • Comparable to the practical difficulty in GIS of showing travel across the date line

  9. Example for higher density at edge neurons (figure from Skupin & Hagelman, 2005) • Census data for TX counties x 3 (1980,1990, 2000) • notice alignment and overplotting along edges

  10. Example for trajectories criss-crossing (figure from Gregg Verutes’ current work) • Movement through knowledge space based on transcribed lectures • Notice long distance moves involving edge neurons (may be real, may be artifact of edges, which is why we need to understand edge effects and whether it helps to counteract them) Graduate Courses Introductory Courses

More Related