apaam/mlpack
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
5,665 Commits 1 Branch 0 Tags
f329dc2b60b18ab83dbf875a67ef8cce856d27a2
Commit Graph
88 Commits
Author SHA1 Message Date
Ryan Curtin f329dc2b60 Document hmm_train executable entirely, and change some outputs from printf to fastlib standards where necessary 2010-05-25 19:50:35 +00:00
Ryan Curtin f539d0979f Rename HMM executables to things that will not potentially conflict with other utilities on a user's system 2010-05-25 19:50:12 +00:00
Ryan Curtin 81bd53e7e7 Add documentation on file formats for HMM utilities 2010-05-25 19:49:52 +00:00
Ryan Curtin a0a801ac3e Document emst; remove non-working --use-thor parameter. Names of some parameters changed for consistency across all MLPACK methods. 2010-05-25 15:46:25 +00:00
Ryan Curtin 766d854026 Update some mistypings in allkfn documentation 2010-05-24 23:07:35 +00:00
Ryan Curtin 2e0b6caaca Properly document the allknn method. 
Previously there was a parameter that allowed selection between single- and dual-tree mode, but I have removed it, defaulting it always to dual-tree mode; I found it was a potentially confusing parameter.
 2010-05-24 23:07:16 +00:00
Ryan Curtin 69bda82205 Properly document allkfn, which was not done at all. Maybe now, people can use it! 2010-05-24 22:47:38 +00:00
Ryan Curtin 40229814a3 Set required parameters and defaults for non-required parameters 2010-05-24 21:58:58 +00:00
tekhnofiend 4f475b8855 updated code so that user can specify number of ICs to find via num_of_IC. Also, fixed bug in very final step where we forgot to transpose unmixing matrix W before multiplying it to data 2010-05-24 21:53:37 +00:00
Ryan Curtin 9713eafc86 Is this a joke? I mean, seriously! allnn ships with no actual way to save the data that it's calculated; I've rewritten so that it might actually be useful to someone, ever 2010-05-24 21:51:23 +00:00
Ryan Curtin e434965a7d Add cheap utility CSV saving method for allnn 2010-05-24 21:50:38 +00:00
Ryan Curtin b1ff13d12a Update documentation so that it actually works, and make sure we tell the user that they can actually give a dataset as a parameter 2010-05-24 21:17:55 +00:00
Ryan Curtin 537c979fe3 Oops, fix -h and -help 2010-05-24 21:06:34 +00:00
Ryan Curtin 49a5b23715 More accurate documentation that meshes better with help screens 2010-05-24 20:57:46 +00:00
Bill March aed59f9ed0 added lower bounds to perm-free, stopped recomputing invalid indices 2010-05-24 20:55:22 +00:00
Ryan Curtin dc9ad359a1 Modify fx output format a little bit so it's a bit easier to understand. Code is ugly but it works... I can't wait to rip this entire system out and replace it with something sensible 2010-05-24 19:49:01 +00:00
Ryan Curtin 9c7407bf34 Make parameter description give --submod/param instead of --param to eliminate confusion 2010-05-21 20:02:19 +00:00
Ryan Curtin 12d0a5ff65 Did not mean to commit this minor change; it turned out to be unnecessary 2010-05-20 16:26:50 +00:00
Ryan Curtin 1b0b1187c3 When a user enters no parameters or -h, fake it so FX thinks the user is asking for help. 
The mere fact that I needed to do THIS to make it work... I have no words.
 2010-05-20 16:23:48 +00:00
Ryan Curtin 55237a2f79 Link against mlpack and Trilinos 2010-05-17 22:46:29 +00:00
Dongryeol Lee f640752d28 Attempt to compile boost unit test. 2010-05-15 15:36:54 +00:00
Ryan Curtin b51f48ee53 Unfortunately, CMake won't work at all if it's being told to configure directories that no longer exist, so I commented out the directories removed in r5575 so it won't freak out anymore. 
Maybe there is a configuration option where it will ignore directories that don't exist?
 2010-05-14 22:15:57 +00:00
Dongryeol Lee ee59960be8 Restarting KDE. 2010-05-14 22:09:18 +00:00
Ryan Curtin ae7c0f1f37 I'm pretty sure the user is aware of what they typed. Don't echo it when they type --help. 2010-05-14 19:25:34 +00:00
Ryan Curtin 77185c8eb1 Change name of variable so -D TRILINOS_LIB_DIR=... can be used 2010-05-11 18:55:31 +00:00
tqlong 4ee7531942 noob test CMAKE 2010-05-09 03:35:28 +00:00
Bill March e5218d5e14 fixed bug in UpdateIndices_, was using wrong index 2010-05-07 17:29:09 +00:00
Bill March e448a0b180 Permutation free version works for just upper bounds as long as I recompute all the distances after each split. Need to work on UpdateIndices_() to figure out why keeping them doesn't work. Add lower bounds after that. 2010-05-06 20:55:54 +00:00
Ryan Curtin e7f06052cf Change TAU to a preprocessor constant to fix odd linker error on Mac (or at least attempt to) 2010-05-06 17:09:50 +00:00
Ryan Curtin 16b6973482 Don't use deprecated values.h; prefer limits.h and float.h instead. 2010-05-04 19:00:34 +00:00
Ryan Curtin 7ae10b6859 Ensure each built library links against BLAS (#28). Apparently this was fine on Linux systems but failed on Macs; hopefully this will solve the issue. 2010-05-03 19:59:06 +00:00
Bill March e86d43b250 updated cmake list 2010-05-03 19:48:06 +00:00
Bill March 21967af0bc wrote permutation free version 2010-05-03 18:51:26 +00:00
Ryan Curtin 93a2e1d6ad Fix for #26; forgot the type specifier after LI (since the LI macro now requires a type after it, so it works for all types) 
#15 (standardized output system using cout) will solve all these silly printf problems.
 2010-04-25 18:13:51 +00:00
Ryan Curtin 949d8116bb posix_fadvise is not defined on all systems 2010-04-24 23:12:51 +00:00
Ryan Curtin 8604c265a4 Do not build mlpack_contrib or any of its pieces by default 2010-04-24 22:33:32 +00:00
Bill March 630d88a0d5 working on multi-bandwidth version 2010-04-21 19:20:03 +00:00
Ryan Curtin 0e9913e5c7 uint -> unsigned int (same thing but 'uint' does not exist on all platforms apparently) 2010-04-21 17:14:00 +00:00
Ryan Curtin 8dfec9dffe Allow selection of SCALE_MASSIVE, SCALE_LARGE, or SCALE_NORMAL in CMake configuration (default to SCALE_NORMAL if not otherwise specified, and check to make sure multiple options are not selected) 2010-04-20 16:55:40 +00:00
Ryan Curtin 8a9f77b3d8 Missed these macros 2010-04-20 16:54:56 +00:00
Ryan Curtin 503651e4af #22: replace lots of preprocessor macros with what they actually are where possible. 
COMPILER_NO_RETURN -> __attribute__((noreturn))
COMPILER_NO_INLINE -> __attribute__((noinline))
COMPILER_PRINTF -> __attribute__((format(printf, X, Y)))
COMPILER_FUNCTIONAL -> __attribute__((const))
COMPILER_DEPRECATED -> __attribute__((deprecated))
COMPILER_DEPRECATED_MSG -> __attribute__((deprecated))
STATIC_CAST(...) -> static_cast<...>(...)
REINTERPRET_CAST(...) -> reinterpret_cast<...>(...)
 2010-04-20 14:58:34 +00:00
Ryan Curtin 2205622dab Incorrect #include line, should use relative paths for fastlib stuff 2010-04-20 14:44:50 +00:00
Ryan Curtin 2f47bc7a6b mmapmm support by default in fastlib means a lot of these methods now compile right 2010-04-15 13:56:12 +00:00
Ryan Curtin 5ead8b1016 Don't use local trilinos headers, use system ones (it's CMake's job to figure out where these are) 2010-04-15 13:55:50 +00:00
Ryan Curtin ab824e4b79 Need to figure out why those two lines have been causing issues 2010-04-14 22:50:19 +00:00
Ryan Curtin 4daef98a6d Add kernel_pca (which required sparse) to default build, modify a few things (ticket #13 is related) so it compiles fine 2010-04-14 22:49:39 +00:00
Ryan Curtin 20ac49f85f Move code that finds trilinos out from sparse CMakeLists 2010-04-14 22:48:32 +00:00
Ryan Curtin d5e7bde7c7 Add mmapmm to fastlib build. A few code changes were necessary stemming from some earlier base/basic_types.h changes. 2010-04-14 21:45:45 +00:00
Ryan Curtin d74c46dc99 Fix nnsvm so it builds correctly (fx_ calls were out of date) and add to CMake build 2010-04-14 21:29:10 +00:00
Ryan Curtin 2bd68567bc Revert 'src' to 'fastlib' so I don't break everything fl-build does. 2010-04-14 15:44:30 +00:00