Ryan Curtin
51e0a9c836
Regularization gets subtracted, not added. Thanks to Sumedh for pointing this
...
out.
2013-11-25 05:59:11 +00:00
Ryan Curtin
6818f76717
Implement FurthestPointDistance() for cover trees.
2013-11-22 18:49:34 +00:00
Ryan Curtin
61e3aa2663
Test BinarySpaceTree::FurthestPointDistance().
2013-11-22 17:02:09 +00:00
Ryan Curtin
be55f84d95
Add FurthestPointDistance() function.
2013-11-22 17:01:59 +00:00
Ryan Curtin
c19d9896ec
Update history file incrementally, because doing this all at once at release
...
time is difficult.
2013-11-21 18:40:14 +00:00
Ryan Curtin
8b4363b5cc
Test constructor that takes an instantiated optimizer.
2013-11-21 18:34:50 +00:00
Ryan Curtin
5b85eb566d
Add a constructor that allows passing an instantiated optimizer.
2013-11-21 18:34:37 +00:00
Ryan Curtin
ac9630f129
Add function to get predictors and responses.
2013-11-21 18:28:01 +00:00
Ryan Curtin
f72372a24a
Add function to get the function that is being optimized.
2013-11-21 18:27:40 +00:00
Ryan Curtin
027850970b
Goodbye, slightly misleading documentation.
2013-11-21 17:58:59 +00:00
Ryan Curtin
1293e2f857
Add tests for actual LogisticRegression object.
2013-11-21 16:57:21 +00:00
Ryan Curtin
871cd988da
Refactor LogisticRegression class. LearnModel() can be private, and we can
...
const-ize a lot of the parameters and functions. Also restore functionality to
Predict(), ComputeError(), and ComputeAccuracy(). This does not ever make a
copy of the predictors matrix and doesn't add a column for the intercept, which
is faster.
2013-11-21 15:40:40 +00:00
Ryan Curtin
2965c9f84e
Don't hold lambda in LogisticRegression because it isn't necessary. Also make
...
predictors and responses const because we don't need to modify them.
2013-11-21 14:23:02 +00:00
Ryan Curtin
fbbee5a8c1
Make predictors and responses const.
2013-11-20 16:44:30 +00:00
Ryan Curtin
2701b42569
Make predictors and responses const.
2013-11-20 16:42:09 +00:00
Ryan Curtin
78933d62dd
Refactor tests so that they do not give the extra column of ones for the
...
intercept term.
2013-11-20 16:37:50 +00:00
Ryan Curtin
ae19acebf1
Refactor functions so that they do not expect the extra column of ones for the
...
intercept term.
2013-11-20 16:37:31 +00:00
Ryan Curtin
cfdfb2fe6d
Don't print "Iteration X of 0."; just "Iteration X".
2013-11-18 20:24:34 +00:00
Ryan Curtin
7aa2f49ea9
Remove memory leak ( #310 ). Thanks to bianjiang for pointing this out.
2013-11-18 18:47:02 +00:00
Ryan Curtin
62108e0e49
And this is why I hate maintaining documentation for things that I don't
...
control; it goes out of date really quickly and people get confused...
2013-11-16 17:06:36 +00:00
Ryan Curtin
766a29b3e2
Add a test for the separable Gradient() function.
2013-11-15 18:22:39 +00:00
Ryan Curtin
18beafc912
Add test for regularization in Gradient().
2013-11-15 18:07:41 +00:00
Ryan Curtin
7267140017
No need to multiply by responses[i] or (1 - responses[i]) because those
...
quantities will always be 1.
2013-11-14 18:12:57 +00:00
Ryan Curtin
7451edc8e0
Test Gradient() in the separable case.
2013-11-14 16:31:48 +00:00
Ryan Curtin
496c18ac52
Update Gradient() for the separable case.
2013-11-14 16:31:41 +00:00
Ryan Curtin
0c17bc6b9b
Test regularization for separable Evaluate().
2013-11-14 16:11:03 +00:00
Ryan Curtin
1793f728e3
Add test for Evaluate(parameters, i) for SGD and related optimizers.
2013-11-14 16:07:22 +00:00
Ryan Curtin
c2be6f0c34
Revamp Evaluate(parameters, i) for SGD and related optimizers. NumFunctions()
...
also needed to be fixed.
2013-11-14 16:07:09 +00:00
Ryan Curtin
432dc71e59
Break Predict() instead of the whole build.
2013-11-13 20:36:11 +00:00
Ryan Curtin
8c33f24ece
Remove getSigmoid() and add class documentation.
2013-11-13 18:28:07 +00:00
Ryan Curtin
0a9b7984bb
Add some comments.
2013-11-13 18:26:56 +00:00
Ryan Curtin
bbecac185e
Make Gradient() const and add comments.
2013-11-13 18:21:52 +00:00
Ryan Curtin
3686feb4fd
Add correct implementation of Gradient() function.
2013-11-13 18:20:38 +00:00
Ryan Curtin
555fb65bb4
Add test for Gradient().
2013-11-13 18:20:27 +00:00
Ryan Curtin
03a801a9ee
Add test for regularization of objective function.
2013-11-13 18:10:51 +00:00
Ryan Curtin
65c6ceaf0d
Add a randomized test for the logistic regression likelihood function that is
...
not just comprised of simple examples.
2013-11-13 18:02:30 +00:00
Ryan Curtin
6c90b28b6f
Oh yeah, this is necessary...
2013-11-13 16:58:25 +00:00
Ryan Curtin
5aece4e4c0
Add test for LogisticRegressionFunction::Evaluate().
2013-11-13 16:42:51 +00:00
Ryan Curtin
c40a059e77
Refactor Evaluate() function so that it works alright, and eliminate unnecessary
...
parameters to the function. Move LogisticRegressionFunction implementation into
a .cpp file because it is not templatized (for now).
2013-11-13 16:42:29 +00:00
Ryan Curtin
91ce8b07ea
LogisticFunction is incorrect; it's actually the objective function for logistic
...
regression, which is not the logistic function. Also remove unnecessary
overloads in LogisticRegressionFunction.
2013-11-11 17:33:28 +00:00
Ryan Curtin
05bc80f9c0
Rename LogisticRegressionFunction to LogisticFunction.
2013-11-11 17:14:30 +00:00
Ryan Curtin
b8d28419fd
Add some comments and fix a little bit of formatting.
2013-11-08 19:23:49 +00:00
Ryan Curtin
76fa71bc39
Initial commit of logistic regression by Sumedh Ghaisas ( #305 ).
2013-11-08 18:52:11 +00:00
Ryan Curtin
f9045e635f
Add logistic regression directory.
2013-11-08 18:50:31 +00:00
Ryan Curtin
e3e80d97cb
Add tests for estimation with the existing model as the initial point.
2013-11-06 03:32:40 +00:00
Ryan Curtin
c4cace4ed8
Allow GMMs to be trained using the existing model as a starting point. This
...
could probably all use some serious refactoring now.
2013-11-06 03:32:23 +00:00
Ryan Curtin
187d49a7d9
Add Epanechnikov kernel for kernel PCA.
2013-10-31 19:41:43 +00:00
Marcus Edel
94714ed0bb
Check the matrix dimension before dropping unnecessary rows.
2013-10-23 11:09:28 +00:00
Ryan Curtin
b1326d1830
Move DTBStat into a separate file.
2013-10-08 19:56:26 +00:00
Ryan Curtin
7c79c1a07d
Widen tolerances slightly.
2013-10-07 17:29:43 +00:00
Ryan Curtin