Skip to content
Commits on Nov 29, 2011
  1. @akashmehra

    ridgeregression write up

    akashmehra committed
Commits on Nov 15, 2011
  1. @akashmehra
  2. @akashmehra
  3. Choosing a better gamma.

    committed
  4. Making Ridge Regression the default method. Data from 1417 trials sho…

    committed
    …w almost
    
    no measurable difference between Ridge Regression and Perceptron.
  5. Adding Ridge Regression. Does not seem to outperform Perceptron. I tried

    committed
    different gamma values [1,2,3,4,5].
Commits on Nov 13, 2011
  1. Use bagging to on perceptron iterations.

    committed
    * Observe that bagging leads to smoother convergence and better performance.
Commits on Nov 12, 2011
  1. Change dater.py to accept dims as in

    committed
      $ dater.py DIMS [FILENAME]
  2. Integrating perceptron Dater and Matchmaker with game architecture.

    committed
    * Use dater.py to generate a weights file as in
      $ dater.py [FILENAME]
    * Use matchmaker.py to play as the matchmaker as in
      $ matchmaker.py HOST PORT DIMS
  3. Tweak learning rate and iterations for better performance and try to …

    committed
    …find more
    
    adversarial Dater preferences by empirical testing.
Commits on Nov 10, 2011
  1. Created alternate methods to find a date, but the original method of …

    committed
    …using the
    
    current hypothesis for the weight vector positive entries is still best.
Commits on Nov 9, 2011
  1. Remove weight constraints after update.

    committed
    * Added notes section at the top of perceptron.py about other things that I
      tried.
  2. Working draft of perceptron.

    committed
Something went wrong with that request. Please try again.