25-08-2017, 09:32 PM
NeuralNetwork.java
NeuralNetwork.docx (Size: 21.08 KB / Downloads: 25)
NeuralNet.java
This class manages most of the neural networking functions for 'Pattern Recognizer' and 'Image Recognizer'. It does not contain any network elements, but operates on matrices provided by client classes.
ImageElements.java
This class describes the individual network element, and is responsible for maintaining its connection weights to the rest of the array.
IOImage.java
This class adds functionality for converting image data to the form, which can be supplied as input to the network and vice versa, for scaling the input image, for adding noise to the input image and for drawing input and output images in input and output fields respectively. Also, this class represents the output field for network operations, storing the main weight matrix. All the network operations are redirected to an instance of the NeuralNet class.
BinaryElements.java
This class describes the individual network element, and is responsible for maintaining its connection weights to the rest of the array.
IOGrid.java
This class is the superclass of InputGrid and OutputGrid, and handles generic element functions.
InputGrid.java
This class is a subclass of IOGrid class. This class adds functionality for following the mouse and activating/deactivating elements accordingly, for scaling input to fill the grid and for adding animated/inanimate noise to the input pattern.
OutputGrid.java
This class is a subclass of IOGrid class. This class represents the output field for network operations, storing the main weight matrix. All the network operations are redirected to an instance of the Neural Net class.
FileActions.java
This class handles file I/O, i.e., storing new user generated patterns to the disk and loading existing patterns from the disk.
PatternTypeFilter.java
This class is an implementation of FileFilter class that filters out all files except for those type extensions that it knows about.
EnterSample.java
This class is used to provide a small component that the user can draw handwritten letters into. This class also contains the routines necessary to crop and down sample the written character. Down sampling is the process of converting a high resolution image to a much lower resolution. The class also displays the down sampled image.
Sample.java
This class adds the sample drawn in the large rectangular area, to an internal data structure.
SampleFileActions.java
This class handles file I/O, i.e., storing new user generated samples to the disk and loading existing samples from the disk.
ListSampleModel.java
This class provides the internal data structure to which the samples are added before they can be trained or saved to the disk.
TrainingSet.java
This class holds a training set for the Kohonen Network. This is usually the set of samples that were inputted by the user.