Unsupervised Learning

Some artificial neural networks, called self-organizing networks, have their connection weights modified via unsupervised learning (Carpenter & Grossberg, 1992; Grossberg, 1980; Grossberg, 1987; Grossberg, 1988; Kohonen, 1977, 1984).  When learning is unsupervised, networks are only provided input patterns.  They are not presented desired outputs that are paired with each input pattern.  In unsupervised learning, each presented pattern causes activity in output units; this activity is often further refined by a winner-take-all competition in which one output unit wins the competition to be paired with the current input pattern.  Once the output unit is selected via internal network dynamics, its connection weights (and possibly the weights of neighboring output units) are updated via a learning rule.

Networks whose connection weights are modified via unsupervised learning develop sensitivity to statistical regularities in the inputs, and organize their output units to reflect these regularities.  For instance, in famous kind of self-organizing network called a Kohonen network (Kohonen, 1984), output units are arranged in a two-dimensional grid.  Unsupervised learning causes the grid to organize itself into a map that reveals the discovered structure of the inputs, where related patterns produce neighboring activity in the output map.  For example, when such networks are presented musical inputs, they often produce output maps that are organized according to the musical circle of fifths (Griffith & Todd, 1999; Todd & Loy, 1991).

References:

1. Carpenter, G. A., & Grossberg, S. (1992). Neural Networks For Vision And Image Processing. Cambridge, MA: MIT Press.
2. Griffith, N., & Todd, P. M. (1999). Musical Networks : Parallel Distributed Perception And Performace. Cambridge, Mass.: MIT Press.
3. Grossberg, S. (1980). How does the brain build a cognitive code?  Psychological review. 87, 1-51.
4. Grossberg, S. (1987).  Competitive learning: From interactive activation to adaptive resonance. Cognitive science, 11, 23-63.
5. Grossberg, S. (1988). Neural Networks And Natural Intelligence. Cambridge, MA: MIT Press.
6. Kohonen, T. (1977). Associative Memory: A System-Theoretical Approach. New
7. Verlag.
8. Todd, P. M., & Loy, D. G. (1991). Music and connectionism. Cambridge, Mass York: Springer-Verlag.
9. Kohonen, T. (1984).  Self-organization and associative memory. New York: Springer-.: MIT Press.

(Added November 2010)