Ryan Curtin
5ec6ce29ac
Require 2.8
2010-04-12 16:53:16 +00:00
Ryan Curtin
7e06699b01
Copy over contrib directory so that I can start adapting build.py to cmake
2010-04-12 16:48:48 +00:00
tekhnofiend
9cc3a0511d
small changes to how files are read in
2010-04-11 18:30:06 +00:00
tekhnofiend
6b4c4ef417
added multivariate dtw that does one global align using Euclidean distance between vectors in R^n
2010-04-11 17:27:17 +00:00
Bill March
abb2f53cb6
actually fixed it now, replaced decrementing a global lower bound with just summing up the bounds as I traverse the new list.
2010-04-08 19:07:48 +00:00
Bill March
465ef3b3fa
fixed bug in hybrid expansion - can't exit checking a node tuple when a pair excludes because the tuple may still have bad symmetry
2010-04-08 18:30:33 +00:00
Bill March
47a41b1454
finished hybrid expansion, added bad symmetry output to node checks
2010-04-07 19:19:41 +00:00
Ryan Curtin
0819d96bb1
Install header files correctly into include/fastlib and include/mlpack.
2010-04-07 16:50:23 +00:00
Ryan Curtin
23fe359c7c
Move header files to correct directory in build (include/mlpack/)
2010-04-06 21:34:55 +00:00
Bill March
8e5694a496
wrote basic hybrid expansion pattern, wrote function to test node lists
2010-04-05 20:42:26 +00:00
Bill March
b4a2009eda
wrote and tested counting function
2010-04-05 18:06:44 +00:00
Ryan Curtin
ed9297f1d9
Update include directories
2010-03-29 23:22:58 +00:00
Ryan Curtin
ee98c6de16
Don't build intermediate libraries, compile everything into libmlpack.a
2010-03-26 21:24:52 +00:00
Ryan Curtin
b396228502
Do not make intermediate libraries; just collect filenames of files to be compiled into libfastlib.a.
2010-03-26 20:10:56 +00:00
houyang
c7a38c968a
Initial release of "regularized risk minimization" project
2010-03-17 20:10:13 +00:00
tekhnofiend
b76c8ee7ed
moved dtw.h implementation into dtw_impl.h
2010-03-15 05:13:11 +00:00
tekhnofiend
c4b937ba12
better organized multivariate dtw and added multivariate dtw to dtw.h
2010-03-15 05:12:53 +00:00
tekhnofiend
e0fff042b8
added straw man - nearest centroid classifier, yields 47 percent accuracy
2010-03-15 01:33:18 +00:00
tekhnofiend
dcb49ff8c1
finished first cut of multivariate dtw
2010-03-15 00:55:25 +00:00
tekhnofiend
929a085dd6
2010-03-15 00:55:12 +00:00
tqlong
79f6b1b088
add cross validation
2010-03-10 21:36:39 +00:00
tqlong
e784b06e6d
code separation into more cc and h files
2010-03-09 19:58:07 +00:00
tqlong
5ac86851d6
code separation into more cc and h files
2010-03-09 19:57:51 +00:00
tqlong
8df8534603
Passive Aggressive
2010-03-05 22:12:08 +00:00
Ryan Curtin
4043770c14
Oops, misspelling. Dangerous because it would compile fine but would not link fx into libfastlib.a.
2010-03-03 18:15:03 +00:00
Ryan Curtin
cad1e31d2c
Revamp cmake structure again. Builds all targets each into their own directory as well as linking into libfastlib.a, making building it all as well as working on individual components simpler. CMakeLists.txt files also commented and clarified to be less confusing.
2010-03-03 17:59:19 +00:00
Bill March
03c11a7da7
wrote depth first recursive code, fixed bug in base case
2010-03-03 01:42:35 +00:00
Ryan Curtin
4a72dd12bf
Revamp cmake:
...
1. Move cmake to top level of fastlib build.
2. Have cmake put libfastlib.a into lib/ (but not the other build components)
3. Have cmake put test executables into bin/ upon build
4. Disable a few things from mlpack build that depend on optimization, until Opt++ problems are solved
Could use some syntax cleanup still.
2010-02-26 22:38:51 +00:00
Ryan Curtin
a64b40d822
One more #include relativity fix
2010-02-26 22:37:12 +00:00
Ryan Curtin
9d01fc94c5
Move CMake utility files to root directory of build (cmake should be run from there).
2010-02-26 22:36:51 +00:00
Ryan Curtin
cf52e842e3
Make #includes relative to fastlib
2010-02-26 22:34:51 +00:00
Ryan Curtin
823721be21
Update #includes to be relative (we should not assume fastlib/ is always in the include path, this keeps it simple).
2010-02-26 22:32:29 +00:00
Ryan Curtin
6a949c02c7
Make clean before distclean
...
(DEPENDS clean did not work right, so this hack will work; after all we already made the assumption we are on a Linux-based system for these commands)
2010-02-26 01:00:01 +00:00
Ryan Curtin
f16b10c9c2
Corrected distclean target for Linux systems. Removes cmake cruft for in-source builds.
2010-02-26 00:49:35 +00:00
Bill March
117299c5bb
fixed a few small errors, made test cases, naive works for 2 and 3 point tests
2010-02-25 19:10:28 +00:00
Bill March
07bdd8deba
and now, to actually add the files
2010-02-24 21:40:06 +00:00
Bill March
08af692788
wrote base case and naive code to use it, few more problems with passing partial tuples around, tests to come
2010-02-24 21:39:36 +00:00
Bill March
37bbaea3f6
n-point correlations, wrote basic kernel function
2010-02-23 21:56:05 +00:00
tekhnofiend
57ec3853da
fixed fucking retarded bug in dtw code that was due to hard coding of time series lengths
2010-02-23 05:22:11 +00:00
tekhnofiend
a7bfa0567f
added dtw visualization tool
2010-02-23 04:08:19 +00:00
tekhnofiend
788fbb4bf5
dtw - final checkthrough
2010-02-23 04:07:57 +00:00
tekhnofiend
3c9a9d472b
changed dtw metric to normalize cost by path_length
2010-02-20 23:00:05 +00:00
tekhnofiend
54ecbaa044
fixed InitRaw member function of ArrayList to allocate memory for capacity, not size
2010-02-11 18:44:20 +00:00
tekhnofiend
b111ab2479
2010-02-11 16:10:47 +00:00
tekhnofiend
f20f3d7615
holy shit dont use arraylist with capacity version of Init
2010-02-11 06:03:26 +00:00
tekhnofiend
cb0494580c
added initial dynamic time warping code - needs debugging
2010-02-11 05:16:41 +00:00
tekhnofiend
07f37f605f
updates to regularized ISM
2010-02-11 05:15:48 +00:00
tekhnofiend
7a7672a82b
updates to spike train dependence
2010-02-11 05:15:34 +00:00
tekhnofiend
485afe35c9
various updates to new_hmm
2010-02-11 05:15:20 +00:00
tekhnofiend
62cbb7f012
update to kernel kmeans
2010-02-11 05:15:08 +00:00
Ryan Curtin