Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Browse files

Merge branch 'master' of https://github.com/nflip/statml

  • Loading branch information...
commit cac5f74940901b4fe22fbe112f2c329d90f91345 2 parents dcdc3c4 + 0bec29d
Marco authored
Showing with 22 additions and 2 deletions.
  1. +22 −2 handin3/README.txt
  2. BIN  handin3/handin3.pdf
View
24 handin3/README.txt
@@ -8,8 +8,28 @@ About the code
All code is implemented in MATLAB.
Neural Network
-----------
-TODO
+--------------
+Run nntrain.m, which
+* creates weight matrices representing eight neural networks,
+* trains them for 25000 batch learning iterations,
+* measures the error to training and test sets every 50 iterations,
+* plots the solutions and the error trajectories,
+* save these plots to solution20ns.eps, solution2.eps, error20ns.eps and error2ns.eps.
+
+The eight neural networks are:
+* four with a hidden layer of 20 neurons,
+* four with two hidden neurons.
+
+Both groups of neural networks feature a training with learning rates of
+* 0.001,
+* 0.0001,
+* 0.00001 and
+* 0.000001.
+
+After running nntrain, these eight weight matrices can be found in WsCells and the error rates in ErrVecTrain and ErrVecTest.
+
+Each major step involves supporting functions. Sometimes alternatives exist. For example, instead of using the function initWsRandNoShortcuts(...) to produce an initial weight matrix without short cuts, initWsRand(...) can be used to get a neural network with short cuts.
+
Support Vector Machines
View
BIN  handin3/handin3.pdf
Binary file not shown
Please sign in to comment.
Something went wrong with that request. Please try again.