HorlogeSkynet/MINDLE
Archived
1
0
Fork 0
Code Issues Pull Requests Releases Activity

Commit Graph

  • b5d3b124ab Update 'README.md' master Samuel FORESTIER 2018-12-22 22:28:17 +01:00
  • 7e800035b4 PEP8 is now respected everywhere 🙇 Samuel FORESTIER 2017-12-23 11:29:58 -05:00
  • 462358f4a0 We actually don't need to specify the encoding for Python 3 programs... ... but the interpreter might be more useful although Samuel FORESTIER 2017-12-23 11:28:01 -05:00
  • 6981f7df35 Improves code style of GradientDescent sources Samuel FORESTIER 2017-12-23 11:26:45 -05:00
  • e514e9b69d Improves code style of Theano sources Samuel FORESTIER 2017-12-23 11:20:07 -05:00
  • 5cf2d72fc2 Removes irrelevant information from README, improves .gitignore and... ... changes target binaries extension (*.bin --> *.out) Samuel FORESTIER 2017-12-23 10:35:13 -05:00
  • 1fd84851a1 Improves code style of TensorFlow code and changes deprecated method Samuel FORESTIER 2017-12-23 09:54:42 -05:00
  • af73a2f405 Improves code style of Museum sources + Removes useless instructions Samuel FORESTIER 2017-12-23 09:39:46 -05:00
  • 28929e6822 Improves code style of MNIST sources #can'tStandTheWarnings Samuel FORESTIER 2017-12-22 22:16:06 -05:00
  • cfb81daa16 Falls back on scalar entities on affineRegression + Minor fixes Samuel FORESTIER 2017-12-22 22:03:47 -05:00
  • 11887ae494 Falls back on scalar entities on linearRegression #weDon'tNeedMatrices Samuel FORESTIER 2017-12-22 16:57:16 -05:00
  • 2df61c4300 Removes meanNormalization for newton and normequ Yann Caumartin 2017-12-21 14:20:28 +01:00
  • 11d1c46b10 Adds normal equation to polynomialRegression Yann Caumartin 2017-12-20 20:21:12 +01:00
  • 72876f992a Adds normal equation to linearRegressionVectors Yann Caumartin 2017-12-20 18:58:08 +01:00
  • 9c612fa8e7 Adds getJ for debugging Yann Caumartin 2017-12-20 16:37:18 +01:00
  • ebf7f8ae0d Adds newton's method in polynomialRegression Yann Caumartin 2017-12-20 16:18:39 +01:00
  • 8e5d42d501 Adds Newton's method in linearRegressionVectors Yann Caumartin 2017-12-19 22:52:36 +01:00
  • daa60ed6d4 Adds required version for sfml & eigen in README Yann Caumartin 2017-12-18 21:33:10 +01:00
  • c049150b1d Adds bin directory in .gitignore Yann Caumartin 2017-12-18 19:02:51 +01:00
  • 20f8a6ae73 Binaries will now be under a bin/ folder to let the project root clean Samuel FORESTIER 2017-12-17 18:58:38 -05:00
  • 9e7d6ca42d Optimizations and code styles Samuel FORESTIER 2017-12-17 18:53:48 -05:00
  • 4eeb2c5f0f We don't need to specify object rules with new versions of make Samuel FORESTIER 2017-12-17 18:41:49 -05:00
  • fffaa4fb54 Changes random engine by C++11's advised one and takes advantage of some Eigen capabilities on affine and linear regressions (to continue) Samuel FORESTIER 2017-12-17 18:30:53 -05:00
  • ad21c3d4c3 Reworks linearRegression with Eigen (+ fix for same previous error) Samuel FORESTIER 2017-12-17 17:12:12 -05:00
  • 660d56baf8 Reworks affineRegression with Eigen and SFML : Working Samuel FORESTIER 2017-12-17 16:58:00 -05:00
  • 912158a2b4 Deletes our Matrix module and reworks the Graphics with Eigen Samuel FORESTIER 2017-12-17 16:57:22 -05:00
  • 116f5d1867 Reworks README and adds a dedicated CONTRIBUTING.md file Samuel FORESTIER 2017-12-17 12:06:51 -05:00
  • 77c9101092 Improves SFML displays by fixing dots radius Samuel FORESTIER 2017-12-17 11:32:36 -05:00
  • 5dac85c8db Just convert each double to long double on graphical codes + Gets rid of some trailing white-spaces Samuel FORESTIER 2017-12-17 11:31:38 -05:00
  • 9dbc5cbd5d Adds eigen dependance in README Yann Caumartin 2017-12-16 11:23:41 +01:00
  • 50a85b90a1 Uses dot() instead of cwiseProduct().sum() Yann Caumartin 2017-12-16 00:11:47 +01:00
  • 04c11ba9ae Converts double to long double Yann Caumartin 2017-12-15 23:58:30 +01:00
  • 73c2e16ed2 Uses eigen in polynomialRegression Yann Caumartin 2017-12-15 22:54:31 +01:00
  • d242c53ecf Uses eigen in linearRegressionVectors Yann Caumartin 2017-12-15 17:20:36 +01:00
  • 63bc2acd92 Fixes typo (INTERNALWIDTH -> INTERVALWIDTH) Yann Caumartin 2017-12-14 17:01:35 +01:00
  • ad053ba2c9 Minor fixes + Change the breaking condition in the training loop ! HorlogeSkynet 2016-11-01 21:42:15 +01:00
  • 047b57eff7 Add LICENCE ! HorlogeSkynet 2016-10-31 13:17:04 +01:00
  • 10ab1eddc5 My personal merge ! HorlogeSkynet 2016-10-31 13:03:37 +01:00
  • 4385661567 Merge branch 'master' of github.com:HorlogeSkynet/MINDLE HorlogeSkynet 2016-10-31 07:50:08 +01:00
  • c06385d6d4 Update README HorlogeSkynet 2016-10-31 07:50:03 +01:00
  • 21fc80548d Introduces mean normalization in polynomialRegression Some things need to be improved Yann Caumartin 2016-10-29 21:35:58 +02:00
  • 63bac56405 Puts the Matrix::operator() in use in linearRegressionVectors & polynomialRegression Yann Caumartin 2016-10-29 20:39:02 +02:00
  • f2390fd541 Implements the operator () in class Matrix Yann Caumartin 2016-10-29 20:07:28 +02:00
  • a3c56960b4 Creates hadamardProduct method for class Matrix Yann Caumartin 2016-10-28 16:15:29 +02:00
  • 8ae9b05f36 Little fixes + Finally add 'polynomialRegression.cpp', which needs to be improved HorlogeSkynet 2016-10-23 15:14:39 +02:00
  • ff3ba8cd25 Remove the -ansi flag in contradiction with the -std=c++11 flag Yann Caumartin 2016-10-22 17:44:07 +02:00
  • 97d649df2a Other little fixes + Improve 'linearRegressionVectors.cpp' HorlogeSkynet 2016-10-20 22:53:04 +02:00
  • b8beb108d3 Merge branch 'master' of github.com:HorlogeSkynet/MINDLE HorlogeSkynet 2016-10-20 22:09:42 +02:00
  • 0fbe4c4b93 Minor fixes and makes the code cleaner HorlogeSkynet 2016-10-20 22:09:36 +02:00
  • 41f298f56d Creation of constructor Matrix(double[],int , int) Yann Caumartin 2016-10-20 22:03:24 +02:00
  • 78ac1489c9 To be improved implementation of the B&B method on multi-parameters examples Yann Caumartin 2016-10-20 20:15:19 +02:00
  • a9ffa62585 Merge HorlogeSkynet 2016-10-20 18:54:59 +02:00
  • 371f2d33d9 Improves Makefile Yann Caumartin 2016-10-20 18:05:34 +02:00
  • 3555f675d0 Creates real() method + corrects Matrix operator* method Yann Caumartin 2016-10-20 14:18:44 +02:00
  • 91cd9b20de Test commit + Rename the old name in 'Makefile' HorlogeSkynet 2016-10-19 20:04:32 +02:00
  • d04d4d453d Update README.md HorlogeSkynet 2016-10-19 19:56:13 +02:00
  • 34d8e85f3d Update README.md HorlogeSkynet 2016-10-18 23:29:16 +02:00
  • 4f24cfd217 Change Makefile and move new modules into a dedicated directory HorlogeSkynet 2016-10-18 23:21:11 +02:00
  • fa4cf37650 Add a new 'graphics' module but still needs to be improved + Fix syntax things + Replaces 'NIL' bad define by 'nullptr' C++11 identifier HorlogeSkynet 2016-10-18 00:28:14 +02:00
  • f60edb630c Adds innerProduct method in class Matrix and fixes constructor Matrix(double**,int,int) Yann Caumartin 2016-10-16 18:57:54 +02:00
  • 8826e304f7 Implement B&B method in affineRegression.cpp and linearRegression.cpp and remove useless loop Yann Caumartin 2016-10-16 16:08:57 +02:00
  • 5ef8324403 Matrix : fixes commutativity problem for operator* + add *=,+=,-= operators + add constructor Matrix(double**,int int) Yann Caumartin 2016-10-15 21:50:55 +02:00
  • f7dc328018 Continues the implementation of the 'Matrix' class but there still remains the reverse multiplication to do... HorlogeSkynet 2016-10-14 21:41:32 +02:00
  • 23e52469e1 Beginning of creation of class Matrix HorlogeSkynet 2016-10-13 19:01:51 +02:00
  • 06fb91f69d Conversion of affineRegression and linearRegression in cpp + add Makefile + modified README HorlogeSkynet 2016-10-13 19:00:59 +02:00
  • 87dc123a37 Fix README HorlogeSkynet 2016-10-08 21:37:41 +02:00
  • a2a4749311 24 hours later, we're leaving TensorFlow + Minor fixes HorlogeSkynet 2016-10-08 21:34:04 +02:00
  • 61b2d61227 Update README.md HorlogeSkynet 2016-10-07 21:35:30 +02:00
  • 5add09013f Renames 'OddsAndEnds' as 'dev_null' + Moves all Theano files in a dedicated directory + Improves and renames 'loadMNIST.py' + Improves README + Initiation with TensorFlow HorlogeSkynet 2016-10-07 21:31:32 +02:00
  • b5ead50ce5 Improve README + Optimising 'learningRateDiscoveringVectors.py' with Theano HorlogeSkynet 2016-09-09 16:45:40 +02:00
  • 4d21b9a59b Minor fixes and improve README HorlogeSkynet 2016-09-09 01:25:09 +02:00
  • 008665f4b6 Minor fixes and merge HorlogeSkynet 2016-07-16 15:51:21 +02:00
  • 596b0b7c13 Minor fixes HorlogeSkynet 2016-07-16 15:37:31 +02:00
  • 5cada265c1 Update TODO Yann Caumartin 2016-07-14 17:24:12 +02:00
  • 106c432886 Very primitive (but successful) implementation of the barzilai&borwein method in learningRateDiscoveringVectors.py Yann Caumartin 2016-07-10 20:24:13 +02:00
  • d044f217fb Separates for all files in LinearRegression Theano's block code in functions stored in computeIterations.py Yann Caumartin 2016-07-10 14:08:44 +02:00
  • ee3101de5b Renaming of some variables and comments on the 'Theano black box' for the average gradient Yann Caumartin 2016-07-09 19:28:35 +02:00
  • be9f2309f3 Merge branch 'master' of github.com:HorlogeSkynet/MachineLearning HorlogeSkynet 2016-07-09 18:55:51 +02:00
  • ffd0b7bb80 Minor fixes HorlogeSkynet 2016-07-09 18:55:15 +02:00
  • 8c2d2668bf Renaming of some variables and comments in 'THEANO BLACK BOX' of linearRegression.py Yann Caumartin 2016-07-09 18:49:10 +02:00
  • 5a2dfe99a2 Merge branch 'master' of github.com:HorlogeSkynet/MachineLearning HorlogeSkynet 2016-07-09 18:48:43 +02:00
  • a5a38b07f5 Improve README + Add copyrights on @authors + "Pythonize" calcul method on linearRegression.py (without gradient) HorlogeSkynet 2016-07-09 18:48:34 +02:00
  • 5fb8ae9c21 Implementation of the 'gradient average' method in linearRegressionVectors.py using Theano Yann Caumartin 2016-07-09 17:34:37 +02:00
  • d5f7968de7 Implementation of the 'gradient average method' in linearRegression.py using Theano Yann Caumartin 2016-07-09 14:33:45 +02:00
  • db544ebb61 Vectorization of testB&B.py Yann Caumartin 2016-07-08 21:47:49 +02:00
  • 4b7a95f17e learningRateNoTheano.py : Making the number of examples and the interval of the input values independent parameters. Adds a costAverage and a gradient Average function. Yann Caumartin 2016-07-07 14:46:32 +02:00
  • 9bad5c9cad Make a main() for all files (following) + Improve README HorlogeSkynet 2016-07-02 21:46:58 +02:00
  • 105a52be0a Makes a main for all file in OddsAndEnds/dev_null and GradientDescent/testB&B.py Yann Caumartin 2016-07-02 21:33:10 +02:00
  • 924383d4ec Adds some comments to the B&B implementation and the linear regression with Cauchy + Reorganizes some file structures + Adds some descriptions at headers + Adds a condition to stop iteration if the learning rate is already null on 'learningRateNoTheano.py' + Rename a file + Finally understood and commented one-hot encoding on 'loadUpdated.py' HorlogeSkynet 2016-07-01 10:19:29 +02:00
  • 4580731788 Add a gradient descent to linearRegressionCauchy.py which uses a cost function computing all the training examples at once Yann Caumartin 2016-06-30 23:25:00 +02:00
  • 0546cca65b Merge of branch master of github Yann Caumartin 2016-06-30 19:38:06 +02:00
  • 81d8225f21 Change type of theta&thetaOld in learningRateNoTheano.py so they doesn't change type automatically at the first iteration + replacement of *= by classic syntax to avoid bugs Yann Caumartin 2016-06-30 19:27:35 +02:00
  • 785e674905 Fix some incomprehensions with Python + Rename a file + Create a new directory which will contains files on the gradient descent + New file on B&B method + New file with Cauchy method on learning rate HorlogeSkynet 2016-06-30 18:59:59 +02:00
  • 8d4285c463 Merge branch 'master' of https://github.com/HorlogeSkynet/MachineLearning Yann Caumartin 2016-06-30 11:22:11 +02:00
  • 1b6a1f13c4 Few corrections to graDescByBoutat.py Yann Caumartin 2016-06-30 11:20:41 +02:00
  • b45319fd10 Merge branch 'master' of github.com:HorlogeSkynet/MachineLearning HorlogeSkynet 2016-06-28 21:05:16 +02:00
  • cad86955ae Minor fixes + Spell correction + Correct gradient descent display HorlogeSkynet 2016-06-28 21:05:03 +02:00
  • 928681e5bc Update TODO Yann Caumartin 2016-06-16 19:43:44 +02:00
  • a18e5f2b1c Finally ! PEP 8 !! + Now functions are closures in graDecent.py HorlogeSkynet 2016-06-16 17:25:29 +02:00
  • 80336ecd98 PEP8 ! HorlogeSkynet 2016-06-16 15:56:59 +02:00