Commit Graph

90 Commits

Author SHA1 Message Date
0452616f2e Added code for basic bidirectional inputs for syllabled letters 2018-03-27 15:28:18 +02:00
54abaf052a Syllabled basic bidirectional ready for test 2018-03-27 15:05:48 +02:00
040969fb09 Updates for bidirectional basic input 2018-03-27 11:55:36 +02:00
cc44e872f6 Prepared bidirectional basic input 2018-03-27 11:41:37 +02:00
92c6368270 Tested bidirectional architectural input 2018-03-27 11:32:20 +02:00
94ce159d44 Test ready bidirectional architectural inputs 2018-03-27 10:35:14 +02:00
9edad0ad07 Commit before major RAM lack update 2018-03-21 11:35:05 +01:00
a316574314 Added functional connected text accentuation for assigning accent location 2017-09-11 16:46:03 +02:00
eb61cb9e74 Added some test results 2017-09-10 17:10:10 +02:00
dbd0b90f92 Added multiple results and error analysis 2017-09-07 19:32:36 +02:00
7180550b0d Added multiple results and created working grid settings and scripts 2017-08-29 20:06:59 +02:00
e79ae5c205 Added some results 2017-08-21 09:30:14 +02:00
68271eebcb Minor changes to workbench 2017-08-19 09:55:25 +02:00
a2fce7c1ae Added num of letters to x_other_features 2017-08-18 19:08:42 +02:00
18348b78fc [MAJOR REFACTOR] Added accent classification (from scratch) and deleted unnecessary y output (output where no accent should be employed) X-es in case of both syllabled inputs have also been changed in similar manner. 2017-07-27 18:20:18 +02:00
83584a0c6f [MAJOR UPGRADE - COMPLETE REFECATOR] Refactored whole prepare_data file into a class, consequently different way of calling functions. Added reverse input option. 2017-07-26 17:03:06 +02:00
76dc71ce55 [MAJOR UPDATE]Added syllabled letters functionallity and added letters attributes to letter accetuation 2017-07-25 09:14:12 +02:00
38797b37d4 [MAJOR UPDATE] Added syllable neural networks 2017-07-21 10:48:50 +02:00
f0d263e429 [MAJOR UPDATE] Changed additional features to version 4, erased unnecessary input letters (unused vowels), split validation data to test data and validation data 2017-07-16 14:29:17 +02:00
0cc949897f Added custom actual accuracy metric which shows how many cases are correctly classified + changed to binary_crossentropy from mse 2017-07-07 16:11:44 +02:00
669aa6bbfd Added fake_epoch for observation 2017-07-07 12:45:47 +02:00
592ad9b26b Simplified nn 2017-07-06 13:41:22 +02:00
cc31b9488b Renamed functional program to cnn 2017-07-06 09:57:04 +02:00
e4e111f029 Added test for y 2017-07-06 09:11:31 +02:00
ca53eedece Changed Sequential functionality to functional API 2017-07-04 13:42:23 +02:00
b84fb13214 merge 2017-07-04 10:41:59 +02:00
cdab668b7c Moved some things 2017-07-04 10:28:27 +02:00
f061573c0b y size changed from 121 to 11 2017-07-02 11:49:41 +02:00
3e76883ce6 Created cnn witch look at aditional features as well 2017-07-01 15:46:51 +02:00
d314a9ee4f Created cnn witch look at aditional features as well 2017-07-01 15:45:46 +02:00
ac8f0057c5 Files renamed, orderd plus ordered prepare_data file 2017-07-01 12:19:09 +02:00
f033638ee9 Changing to python3 2017-06-30 13:43:35 +02:00
4d27925b87 Changed to python3 2017-06-30 13:18:48 +02:00
719cec6d29 Added some modularity and X_other_features generation 2017-06-28 16:04:04 +02:00
3c9edd5d1c Managing validate data to counter overfitting 2017-06-27 11:40:56 +02:00
0e3dd4f88c Added parts of per vowel accetuation 2017-06-24 12:35:23 +02:00
dfe4b9a362 Added creation of inputs per vowels 2017-06-23 11:49:21 +02:00
2c16f10e13 Created character_based_cnn 2017-06-20 12:50:34 +02:00
658c55d280 More tutorial 2017-04-13 11:10:23 +02:00
9455f26423 Initial commit 2017-02-10 10:31:17 +01:00