Conclusion

• Take Home Points
  1. SOMs are a series of nodes, typically in Euclidean space.
  2. They are trained by an unsupervised system, typically using the Mexican Hat function.
  3. They are good for clustering.
  4. Self organisation is very powerful, but poorly understood.
• The Canonical Book for SOMs is Tuevo Kohonen's Self-Organizing Maps.
• Reading: for this week is the self-organizing map wiki.
• Reading: For next week is Russell and Norvig 461-463 (the Semantic Nets section 5.1 of the Knoweldge Representation Chapter). The Semantic Net wiki is also pretty good.