# publiclaumann/statml

### Subversion checkout URL

You can clone with HTTPS or Subversion.

Merged
merged 15 commits into from over 2 years ago
+49 4

### 3 participants

NN part still needs to be done

 nflip add findMods: create a vector of modifiable indices of weight matrix ae1f216 nflip modify Er.m: optimization 6acba54 nflip modify estimatePds: optimization by using modifiable-vector instead o… …f iterating through all 5079286 nflip modify Er.m: change length to size for matrix size 1064e24 nflip result of computation: 50 * 500 trainings cce18bb nflip add initWsRandNoShort: a weight init without shortcuts from input to … …output 6828561 Marco Git is fucking with me 5f43f40 Marco Started README, still incomplete 97eba2d nflip Merge remote branch 'marco/master' 54dbbd5 nflip add nntrain information to README.txt 9f56d6d nflip retex handin3.pdf for eps graphics 304afca nflip modify README.txt: extended NN 0bec29d Marco Fixed error in regularization and README dcdc3c4 Marco Merge branch 'master' of https://github.com/nflip/statml cac5f74 Marco fixed typo e38e32a
merged commit e23cab5 into from
closed this

Showing 15 unique commits by 3 authors.

Mar 20, 2012
add findMods: create a vector of modifiable indices of weight matrix ae1f216
modify Er.m: optimization 6acba54
modify estimatePds: optimization by using modifiable-vector instead o…
…f iterating through all
5079286
modify Er.m: change length to size for matrix size 1064e24
result of computation: 50 * 500 trainings cce18bb
add initWsRandNoShort: a weight init without shortcuts from input to …
…output
6828561
Git is fucking with me 5f43f40
Started README, still incomplete 97eba2d
Merge remote branch 'marco/master' 54dbbd5
add nntrain information to README.txt 9f56d6d
retex handin3.pdf for eps graphics 304afca
modify README.txt: extended NN 0bec29d
Fixed error in regularization and README dcdc3c4
Merge branch 'master' of https://github.com/nflip/statml cac5f74
fixed typo e38e32a
 @@ -253,7 +253,7 @@ \subsubsection{Effect of the regularization parameter} 253 253 254 254 The file \texttt{regularization.m} performs the outlined procedure, by first training the SVM model using the values for $C$ and $\gamma$ found during model selection. Then it trains to other models, one in which $C$ is multiplied by a hundred and one in which we divide $C$ by 100. 255 255 256 -The most notable change is in the number of support vectors. There's a total of 93 support vectors for the original'' value of $C$---87 of which are bounded. When $C$ is a hundred times larger, the number of support vectors drop to just 19, all of which are free. Conversely, when dividing $C$ by a hundred we get an increase in the number of support vectors to 199, but again all of them are free. 256 +The most notable change is in the number of support vectors. There's a total of 93 support vectors for the original'' value of $C$---87 of which are bounded. When $C$ is a hundred times larger, the number of support vectors drop to just 19, 14 of which are bounded. Conversely, when dividing $C$ by a hundred we get an increase in the number of support vectors to 199, and again most of them (195) are bounded. 257 257 258 258 \subsubsection{Scaling behaviour} 259 259