select_feature_set_trainf_mlp — Selects an optimal combination of features to classify OCR data.
select_feature_set_trainf_mlp selects an optimal combination of
features, to classify the OCR data given in the training file
TrainingFile with a multilayer perceptron,
for details see
Possible features are all OCR features listed and explained in
create_ocr_class_mlp. All candidates which should be tested can be
FeatureList. A subset of these features is
returned as selected features in
select_feature_set_trainf_mlp is specialized on OCR problems and
only supports the features in the list mentioned before.
In order to use other features, please use the more general operator
The selection method
SelectionMethod is either a greedy search 'greedy'
(iteratively add the feature with highest gain)
or the dynamically oscillating search 'greedy_oscillating'
(add the feature with highest gain and test then if any of the already added
features can be left out without great loss).
The method 'greedy' is generally preferable, since it is faster.
Only in cases when a large training set is available
the method 'greedy_oscillating' might return better results.
The optimization criterion is the classification rate of a two-fold
cross-validation of the training data. The best achieved value
is returned in
to adapt the setting of the number of hidden neurons in the MLP with
'num_hidden'. The default value is 80, a higher value
leads to longer training times but might lead to a more expressive
This operator may take considerable time, depending on the size of the data set in the training file, and the number of features.
Please note, that this operator should not be called, if only a small
set of training data is available. Due to the risk of overfitting the
select_feature_set_trainf_mlp may deliver a classifier with
a very high score. However, the classifier may perform poorly when tested.
This operator returns a handle. Note that the state of an instance of this handle type may be changed by specific operators even though the handle is used as an input parameter by those operators.
Names of the training files.
Default value: ''
File extension: .
List of features that should be considered for selection.
Default value: ['zoom_factor','ratio','width','height','foreground','foreground_grid_9','foreground_grid_16','anisometry','compactness','convexity','moments_region_2nd_invar','moments_region_2nd_rel_invar','moments_region_3rd_invar','moments_central','phi','num_connect','num_holes','projection_horizontal','projection_vertical','projection_horizontal_invar','projection_vertical_invar','chord_histo','num_runs','pixel','pixel_invar','pixel_binary','gradient_8dir','cooc','moments_gray_plane']
List of values: 'anisometry', 'chord_histo', 'compactness', 'convexity', 'cooc', 'default', 'foreground', 'foreground_grid_16', 'foreground_grid_9', 'gradient_8dir', 'height', 'moments_central', 'moments_gray_plane', 'moments_region_2nd_invar', 'moments_region_2nd_rel_invar', 'moments_region_3rd_invar', 'num_connect', 'num_holes', 'num_runs', 'phi', 'pixel', 'pixel_binary', 'pixel_invar', 'projection_horizontal', 'projection_horizontal_invar', 'projection_vertical', 'projection_vertical_invar', 'ratio', 'width', 'zoom_factor'
Method to perform the selection.
Default value: 'greedy'
List of values: 'greedy', 'greedy_oscillating'
Width of the rectangle to which the gray values of the segmented character are zoomed.
Default value: 15
Height of the rectangle to which the gray values of the segmented character are zoomed.
Default value: 16
Names of generic parameters to configure the selection process and the classifier.
Default value: 
List of values: 'nu'
→(real / integer / string)
Values of generic parameters to configure the selection process and the classifier.
Default value: 
Suggested values: '0.1'
Trained OCR-MLP classifier.
Selected feature set, contains only entries from
Achieved score using tow-fold cross-validation.
If the parameters are valid, the operator
select_feature_set_trainf_mlp returns the value 2 (H_MSG_TRUE). If necessary,
an exception is raised.