#Test steps:
- Implement file IFileSystem and IFileSystemItem
Fill followed properties in config.properties
filesystem class name
configuration.filesystem.class=
contructor objects for file system constructor
configuration.filesystem.constructor.args=
class types for file system constructor
configuration.filesystem.constructor.args.types= - Implement ISignal, ISignalProcessor, ISignalSerializer,INConnection and IWeight.
- Create HashMap<Class<? extends ISignal>, ISerializer<? extends ISignal, String>>
serialize to bytes and save to file. You can use for this purpose ByteSaverHelper. - Put the result from step 3 to config.properties
signal serializers in byte view
configuration.serializers= - Implement IStudyingAlgorithm and IStudyingRequest.
Serialize Studying Algorithm to bytes and save to file. - Put the result from step 5 to config.properties
Studying algo in byte view
configuration.studyingalgo= - Add input for first layer and create input folders for other layers
it should looks like $inputFolder/$layerId/file
file format {"inputs":[{"neuronId":"$longID","signal":{"className":"$signalClass",...}},...]}
Note: ISignalSerializer which You have implemented in step 2 should serialize/deserialize ISignal with field className Result object for studying should be stored her as bytes $inputFolder/result use ByteSaverHelper - Put path to input dir to the config.properties
path to input folder
configuration.input.path= - create layers configuration in such format
{"layerID":"$layerID", "layerSize":"$size","neurons":"$serializedToBytesArrayOfNeurons"}
note: You must start layer from 0 and increment by 1 i.e input layer is 0 - Place layers configuration in such way $layersPath/$fileNameAsLayerID
- Specify path to layer configuration in config.properties
path to neuron net configuration
configuration.input.layermeta=
Fill free to contact me You will find errors/bugs or will need assist.