http://en.wikipedia.org/wiki/Supervised_learning...

http://en.wikipedia.org/wiki/Supervised_learning
http://en.wikipedia.org/wiki/Unsupervised_learning

In General Terms:
With Supervised Learning, the data supplied to the Artificial Neural Network (ANN) has been broken up into different sets, with an indicator telling the ANN which set each sample of data belongs to. When the Neural Network is then fed data to evaluate, it works out which set the data belongs to.

So, feed the ANN samples of Music, Rock, Jazz, Elevator, etc, with Labels stating which Genre the music belongs to. Then supply music WITHOUT Labels and the ANN will tell you which Genre of music they belong to.

With Unsupervised learning, the data supplied to the Artificial Neural Network (ANN) has NOT been broken up into different sets. The ANN works out its own breakup of the data into different sets. When the Neural Network is then fed data to evaluate, it works out which of its OWN sets the data belongs to. The ANN can find Classifications that Humans have missed which may have worth as separate Classifications

So, feed the ANN samples of Music, Rock, Jazz, Elevator, etc, with NO indicator showing that music is even considered as belonging to different Genres. The ANN will work out its own Genre Classifications for the music. Then supply music to the ANN and it will tell you which of its OWN Genre Classification the music belongs to