 vasiloglou
|
dc9aefc2db
|
close to automatic tuning
|
2008-03-24 20:48:41 +00:00 |
|
|
vasiloglou
|
f849e3f7cd
|
fixed a couple of bugs in the max furthest
|
2008-03-21 21:37:21 +00:00 |
|
|
vasiloglou
|
45d581b7a4
|
Now the submodule works better
|
2008-03-20 20:35:15 +00:00 |
|
|
vasiloglou
|
27d858a230
|
Eventually it worked without warnings
|
2008-03-20 19:50:12 +00:00 |
|
|
vasiloglou
|
9751f60e0a
|
Trying to solve the submodule issue
|
2008-03-20 17:34:19 +00:00 |
|
|
vasiloglou
|
c576fb23a0
|
There was a bug with the num_iterations that wasn't initialized
|
2008-03-20 17:21:23 +00:00 |
|
|
vasiloglou
|
2d59fb7d8a
|
I need to get rid of some warnings, it is running
|
2008-03-20 17:09:58 +00:00 |
|
|
vasiloglou
|
00cd177dfc
|
compiles
|
2008-03-20 15:56:45 +00:00 |
|
|
vasiloglou
|
c3698de85b
|
Added a recorder for the final results
Main function for mvu
|
2008-03-20 05:45:33 +00:00 |
|
|
vasiloglou
|
077762e294
|
more
|
2008-03-19 04:15:07 +00:00 |
|
|
vasiloglou
|
57fdd2df03
|
Added some extra files
|
2008-03-19 03:48:39 +00:00 |
|
|
vasiloglou
|
5970355d5f
|
Implemented the member functions
|
2008-03-19 03:07:16 +00:00 |
|
|
vasiloglou
|
3901a8d029
|
Some minor repairs
|
2008-03-19 00:16:58 +00:00 |
|
|
vasiloglou
|
d89b20d088
|
fisrt check in
|
2008-03-19 00:15:23 +00:00 |
|
|
vasiloglou
|
acc3b2b164
|
several bugs fixed
|
2008-03-13 18:46:15 +00:00 |
|
|
vasiloglou
|
ec015dcf09
|
some minor modifications, tests seem to run ok
|
2008-03-13 18:45:55 +00:00 |
|
|
vasiloglou
|
eb82302da2
|
It is working
|
2008-03-13 13:15:25 +00:00 |
|
|
vasiloglou
|
3b80096ffb
|
test seems to pass, we have to check for the validity of the results
|
2008-03-13 05:22:36 +00:00 |
|
|
vasiloglou
|
cade017c70
|
Remove some binary files tha accidently were added
|
2008-03-13 02:58:28 +00:00 |
|
|
vasiloglou
|
cddb821d4f
|
Compiling
|
2008-03-13 02:57:05 +00:00 |
|
|
Dongryeol Lee
|
2cc2d27794
|
Fixed the compilation error
|
2008-03-11 02:09:52 +00:00 |
|
|
Dongryeol Lee
|
d02458f3b2
|
The driver application now reads in a dataset of short ints for demonstration.
|
2008-03-11 01:50:20 +00:00 |
|
|
rriegel
|
4566179385
|
fixed test suite, fixed arraylist ambiguity
|
2008-03-11 01:46:08 +00:00 |
|
|
Bill March
|
e55c93e7cf
|
stuff
|
2008-03-11 01:03:17 +00:00 |
|
|
Dongryeol Lee
|
900ca48466
|
Fastlib tree now is OT-compliant, meaning it can be serialized/unserialized. However, it still does not admit a bounding box of general primitive type (for example, a short-int based bounding box rather than a double-based bounding box), which would involve making changes to bounds.h, math/math.h:DRange class.
|
2008-03-09 00:37:59 +00:00 |
|
|
Dongryeol Lee
|
2ddadcc3f2
|
Separated out the naive algorithm into a separate file. Now algorithms are templatized for different primitive types
|
2008-03-09 00:32:12 +00:00 |
|
|
rriegel
|
7991f3aba7
|
templatized matrices
|
2008-03-07 03:59:59 +00:00 |
|
|
rriegel
|
d918575805
|
some memory optimizations
|
2008-03-07 03:58:39 +00:00 |
|
|
Bill March
|
bb43bb75bc
|
code hooray
|
2008-03-05 18:55:08 +00:00 |
|
|
Dongryeol Lee
|
2aace707ca
|
Numerical stability fix
|
2008-03-05 01:39:30 +00:00 |
|
|
Dongryeol Lee
|
f703694b81
|
More fixes in CG loop regarding the updating the solutions
|
2008-03-05 01:17:54 +00:00 |
|
|
Dongryeol Lee
|
0d6b734001
|
Bug fix in the Krylov loop - changed a break statement in the for-loop that iterated over each query to a continue statement
|
2008-03-05 01:05:31 +00:00 |
|
|
Dongryeol Lee
|
d05803cc18
|
indentation fix in multi_conjugate_gradient.h
|
2008-03-05 00:54:32 +00:00 |
|
|
Dongryeol Lee
|
1e64b08e0c
|
More parameter tweaking in the CG loop
|
2008-03-05 00:43:03 +00:00 |
|
|
Dongryeol Lee
|
d639f4894f
|
Fixed the normalization constant handling for query influence values
|
2008-03-05 00:35:02 +00:00 |
|
|
Dongryeol Lee
|
89b57d0c37
|
I need to fix the Krylov loop to make sure all three system solutions have converged before a query exists the Krylov loop.
|
2008-03-05 00:16:42 +00:00 |
|
|
Dongryeol Lee
|
5b1d0cbbcd
|
Edited the krylov-based regression code for numerical stability
|
2008-03-03 21:31:07 +00:00 |
|
|
Dongryeol Lee
|
255dfcebe5
|
Tweaked the termination condition in the CG algorithm
|
2008-03-03 21:21:56 +00:00 |
|
|
Dongryeol Lee
|
bd9dfdcf9e
|
Last generation of the Krylov subspace based method has been caught - the bug was due to reusing the same dual-tree computation function across three different phases (some phases were required to use squared kernel values, rather than the regular values)
|
2008-03-03 20:52:57 +00:00 |
|
|
Dongryeol Lee
|
221c1ce793
|
Bug fix in the dense version: same problem with the naive version since I forgot to divided by the normalization constant for the influence values
|
2008-03-03 04:28:20 +00:00 |
|
|
Dongryeol Lee
|
6a2a91e22b
|
Minor fixes in the user-level function Compute
|
2008-03-03 01:38:23 +00:00 |
|
|
Dongryeol Lee
|
c69e48ef2b
|
Some work for setting up the experiment for the dense version of the algorithm has been completed. I now need to look at the krylov subspace version for correctness since I do some fancy stuff for the normalization constants
|
2008-03-02 22:13:27 +00:00 |
|
|
Dongryeol Lee
|
d315ca87b5
|
Krylov subspace based LPR now computes full information required for crossvalidation: confidence bands, regression estimate, leave-one-out regression estimate, influence value, and magnitude of its weight diagram vector in the hat matrix
|
2008-03-02 19:27:39 +00:00 |
|
|
Dongryeol Lee
|
f39afd4da8
|
Fix in the compute function for naive code which contained erroneous extra argument
|
2008-03-02 06:11:51 +00:00 |
|
|
Dongryeol Lee
|
b8f8db9bf5
|
Now shuffles the values for weight diagram and influence values for Krylov version
|
2008-03-02 04:59:18 +00:00 |
|
|
Dongryeol Lee
|
f6dba35e8b
|
Fixed a bug in the dense version of the algorithm which did not shuffle the computed magnitude of weight diagrams and influence values accordingly due to the query tree construction; also there was a logic mistake in the design of the code which assumed that under variable-bandwidth case, the epanechnikov moment computed for the denominator matrix could be used for the weight diagram matrix, but this is not true since the effective normalization constant for weight diagram matrix requires squared normalization constants
|
2008-03-02 04:39:00 +00:00 |
|
|
Dongryeol Lee
|
28e2e2b364
|
Now computes the influence values for each reference point, need to add in the final dual-tree pass for computing the magnitude of each weight diagram vector for each reference point to construct the confidence interval
|
2008-03-01 21:51:31 +00:00 |
|
|
Dongryeol Lee
|
fe05acf2ed
|
Fixed the compilation error for general kd-tree code
|
2008-03-01 21:40:40 +00:00 |
|
|
Dongryeol Lee
|
e5e724a387
|
Now precomputes query point expansions before finalizing regression estimates, since they are needed in computing the confidence intervals (need to be passed to the Krylov solver). The Krylov solver needs to be modified, plus one final dual-tree pass over the computed solution vectors need to be implemented
|
2008-03-01 05:22:57 +00:00 |
|
|
Dongryeol Lee
|
6de282c263
|
Added the leave-one-out regression estimates for the reference set training stage
|
2008-03-01 04:34:38 +00:00 |
|