apaam/mlpack
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
14,390 Commits 1 Branch 0 Tags
b8bb079a5feb7e8d95cdbd6094bc2fa28b5e1971
Commit Graph
14390 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Ryan Curtin b8bb079a5f Refactor RefinedStart to give centroids not assignments. 2016-04-12 14:43:52 +00:00
Ryan Curtin a38608bea2 Add SampleInitialization and different type of initialization. 2016-04-12 14:43:52 +00:00
Ryan Curtin cb3c489bee Don't set the random seed! That is done at higher levels. 2016-04-12 14:43:52 +00:00
Ryan Curtin 243ace7886 Document the DatasetInfo class more comprehensively. 2016-04-12 14:43:52 +00:00
Ryan Curtin 6ccf865445 Add a tutorial for the formats. 2016-04-12 14:43:52 +00:00
Ryan Curtin a955ee174f Clarify section name. 2016-04-12 14:43:52 +00:00
Ryan Curtin eeba6bdc50 Merge pull request #612 from stereomatchingkiss/refine_log_softmax 
use constexpr to replace static const
 2016-04-12 09:26:40 -04:00
stereomatchingkiss 63be94ebfa use constexpr to replace static const 2016-04-12 19:01:18 +08:00
Ryan Curtin 78f6ac2e03 Merge pull request #611 from ranjan1990/SGD 
little change in SGDTestFunction::Gradient()
 2016-04-11 12:13:47 -07:00
ranjan1990 d1c2ae0b09 branch conflict resolved 2016-04-12 00:36:16 +05:30
ranjan1990 3f2d6ab56d name added 2016-04-12 00:02:59 +05:30
ranjan1990 7e1e1deb04 little change 2016-04-11 22:15:57 +05:30
Marcus Edel 460e326cb2 Add LogSoftmaxLayer class; the log softmax loss layer computes the multinomial logistic loss of the softmax of its inputs. 2016-04-11 16:09:17 +02:00
Marcus Edel c0dbd32ac0 Add ConstantLayer class that outputs a constant value given any input. 2016-04-11 16:07:46 +02:00
Marcus Edel fb3994c48e Add option to set the input parameter when calculating the gradient. 2016-04-10 15:49:35 +02:00
Marcus Edel 592fdcda15 Reorder the parameter of the DropConnectLayer to avoid reorder warning. 2016-04-09 18:08:09 +02:00
Marcus Edel 0e8d776e03 Make sure, the gradient isn't empty, before propagating the gradient parameter through the network. 2016-04-09 13:36:50 +02:00
Marcus Edel f4b3464fce Add confidence parameter to enable custom class transformation. 2016-04-09 13:31:20 +02:00
Marcus Edel 07289ba8d6 Add test case for the layer and network weights function that uses a given initialization rule to initialize the weights. 2016-04-09 13:31:20 +02:00
Marcus Edel b33964b931 Add auxiliary function to set the network weights using a given initialize rule. 2016-04-09 13:31:20 +02:00
Marcus Edel f941837e68 Add test case for the layer input size and network input size function. 2016-04-09 13:31:20 +02:00
Marcus Edel d0708a5f6d Add function to get the input size of a given network. 2016-04-09 13:31:20 +02:00
Ryan Curtin ba826b1959 Make cleaner Armadillo expression. 2016-04-08 20:51:15 -04:00
Ryan Curtin bc6b9e7476 Style fixes for LSH test. 2016-04-08 20:37:52 -04:00
Ryan Curtin f0675d7789 Merge pull request #605 from mentekid/issue261/LSH_test 
Issue261/lsh test
 2016-04-06 17:56:57 -04:00
Yannis Mentekidis 9ef75a49d4 Whitespace fixes 2016-04-06 09:55:56 +01:00
Yannis Mentekidis f229e4e755 Separate tests into Boost test cases 2016-04-06 09:52:59 +01:00
Yannis Mentekidis c8385668ad Style Fixes In lsh_test 2016-04-05 23:02:10 +01:00
Marcus Edel 5bc514c122 As pointed out in the program info; the reference set will be used as both the reference and query set if no separate query set is specified. 2016-04-05 21:47:41 +02:00
Yannis Mentekidis 9561fe2317 Add Yannis Mentekidis to contributors 2016-04-05 19:18:09 +01:00
Yannis Mentekidis 18c925ad01 Style changes 2016-04-05 19:17:27 +01:00
Yannis Mentekidis b0e5d740b4 Fix failsafe for numTables recall test 2016-04-05 16:41:35 +01:00
Yannis Mentekidis 93a42c88ae Merge branch 'master' of https://github.com/mlpack/mlpack into issue261/LSH_test 2016-04-05 11:43:40 +01:00
Yannis Mentekidis 289782e7d4 Style Fixes 2016-04-05 11:43:20 +01:00
Yannis Mentekidis bced046b89 Add failsafe to numProj test of LSHSearch 2016-04-05 11:39:54 +01:00
Yannis Mentekidis 1708dfdc24 Add 3 more recall-based tests for LSHSearch 2016-04-05 11:38:39 +01:00
Ryan Curtin 9f72e89dda Fix unnecessarily long line. 2016-04-04 23:29:20 -04:00
Ryan Curtin e14c246389 Document cosine tree fix. 2016-04-04 22:53:50 -04:00
Ryan Curtin 6a70cfcf8f Minor style fixes. 2016-04-04 22:45:13 -04:00
Ryan Curtin fb75f38b6a Merge pull request #581 from lozhnikov/cosinetree 
Fix bug #358 (CosineTreeTest/CosineNodeCosineSplit test sometimes fails, only on i386)
 2016-04-04 22:42:36 -04:00
marcus 66cb285dc7 Divide the root square of the second decay value by the first decay value; clarification of the beta parameter. 2016-04-01 18:17:57 +02:00
marcus 6359987ecb Add simple test that to ensure the batch size isn't zero. 2016-04-01 17:57:25 +02:00
marcus 68e2cbf394 Ensure the last batch size isn't zero, to avoid division by zero before updating, when using a batchsize that fulfilled the constraint: (numFunctions % batchSize) == 1. 2016-04-01 17:57:25 +02:00
Yannis Mentekidis 799cafee84 Add testing for LSHSearch and refactors code 2016-03-31 15:22:23 +03:00
Mikhail Lozhnikov 1e1bc6fbf9 Use BOOST_REQUIRE_* instead of BOOST_CHECK_* 2016-03-28 18:23:19 +03:00
Yannis Mentekidis 0031a82e40 Adds two tests for LSHSearch 2016-03-26 18:44:04 +02:00
Ryan Curtin 9e4e126589 Merge pull request #592 from kno10/patch-1 
Bug fix for Elkan?
 2016-03-25 23:59:24 -04:00
Ryan Curtin a0f1dd5632 Fix dual-tree k-means runtime bug. 
The residue was being calculated incorrectly, and as a result it would run until
the maximum number of iterations.  Thanks to Erich Schubert for bringing this
one up and pointing it out.

The fascinating thing is that this did not trigger any test failures, because it
still converged to the correct results---it just took way longer than it should
have.
 2016-03-25 21:17:32 -04:00
Ryan Curtin c14e81a563 Fix link. 2016-03-25 18:00:34 -07:00
Erich Schubert 04b0c0867f Minimal fix for bug in Elkan implementation. 
Kill only one cluster, not all.
 2016-03-25 22:50:01 +01:00