Merge branch 'evaluation' |
|
More...
|
about 10 years ago
|
Log the basic performance measures in the log, as well as dumping the Confusion Matrix |
|
More...
|
about 10 years ago
|
Merge branch 'evaluation' |
|
More...
|
about 10 years ago
|
Refactored code to teach neural network into separate method |
|
More...
|
about 10 years ago
|
Refactoring: Made a function to evaluate all neural networks i'm interested in |
|
More...
|
about 10 years ago
|
Refactored the process of loading the char data for a data set |
|
More...
|
about 10 years ago
|
Refactored code to extract just the character from an image to seperate methods in it's own class |
|
More...
|
about 10 years ago
|
Bitmap to double[,] |
|
More...
|
about 10 years ago
|
Use cross-validation data when training the neural network to detect over-learning and if the performance has plateaued |
|
More...
|
about 10 years ago
|
Basics of a working Evaluation system. Code in Quantative.Evaluation required pretty major refactoring |
|
More...
|
over 10 years ago
|
Initialise Log |
|
More...
|
over 10 years ago
|
Log system |
|
More...
|
over 10 years ago
|
Stubbed Quantitative Evaluation system |
|
More...
|
over 10 years ago
|
Refactored all Exceptions inside SharedHelpers to be in the SharedHelpers.Exceptions namespace |
|
More...
|
over 10 years ago
|
MainForm: for some reason this never got committed when it was originally written |
|
More...
|
over 10 years ago
|
Removed reference to DevCharClassification from the solution (not in this branch) |
|
More...
|
over 10 years ago
|
Merged DevCharClassification, leaving the actual DevCharClassification off of the main branch |
|
More...
|
over 10 years ago
|
Refactor: class name consistency |
|
More...
|
over 10 years ago
|
Merge branch 'master' |
|
More...
|
over 10 years ago
|
Don't check for sign of difference in Area, because now the points are ordered clockwise, sumB will always be greater than (or equal to) sumA |
|
More...
|
over 10 years ago
|
Test comments moved for consistency |
|
More...
|
over 10 years ago
|
new method for sorting points into a clockwise rotation, allowing for the contained area of an unordered list of points to be calculated correctly |
|
More...
|
over 10 years ago
|
Tweaked caharcter resolutions to minimise misclassifications |
|
More...
|
over 10 years ago
|
Output images of the misclassified training data so you can see what the system is stuggling to learn |
|
More...
|
over 10 years ago
|
Bug fix: copy paste - now evaluates the evaluation data rather than doing the training data twice |
|
More...
|
over 10 years ago
|
Evaluate the trained network on the evaluation data |
|
More...
|
over 10 years ago
|
Refactored some code ready for evaluation on the evaluation data |
|
More...
|
over 10 years ago
|
Network runs and tells you the number of characters from the training data that were misclassified by the trained classifier |
|
More...
|
over 10 years ago
|
Converted methods using float[] for the Neural network to use double[] instead (as the newer versions of AForge.NET do) |
|
More...
|
over 10 years ago
|
Logic to traing the neural network implemented |
|
More...
|
over 10 years ago
|