nov 19

homework

• 18.3 - decision tree for traffic light
• 20.11 - XOR neural net
• 20.13 - attempt to train neural net with no hidden layers on four input parity function : what happens?
• final projects

markov etc - overview of next topics

• see Dec 3 assignment
• Walk through basic ideas of
  • markov chain
  • hidden markov model
    • chap 15
    • wikipedia: Hidden Markov Model
    • U Leeds HMM tutorial
    • http://ghmm.org/ : GNU HMMM software; C library with Python bindings
  • connection to finite state machine
  • connection to formal grammars
• example : speech recognition, pg 568 and following
• two different but closely related problems :
  • generating "intelligent" patterns
  • analyzing (parsing) patterns "intelligently"
• related CS topics
  • BNF : http://en.wikipedia.org/wiki/Backus-Naur_form
  • yacc, bison, etc : http://en.wikipedia.org/wiki/Parser_generator
• Abe: http://en.wikipedia.org/wiki/Algorithmic_music
• Other generators
  • Scigen
    • http://pdos.csail.mit.edu/scigen/
    • http://cs.marlboro.edu/code/perl/SCIgen/scigen/
  • Context Free Design Grammar
    • http://www.chriscoyne.com/cfdg/