Class OctonionNeuralNetwork<T>
- Namespace
- AiDotNet.NeuralNetworks
- Assembly
- AiDotNet.dll
Represents an Octonion-valued Neural Network for processing data in 8-dimensional hypercomplex space.
public class OctonionNeuralNetwork<T> : NeuralNetworkBase<T>, INeuralNetworkModel<T>, INeuralNetwork<T>, IFullModel<T, Tensor<T>, Tensor<T>>, IModel<Tensor<T>, Tensor<T>, ModelMetadata<T>>, IModelSerializer, ICheckpointableModel, IParameterizable<T, Tensor<T>, Tensor<T>>, IFeatureAware, IFeatureImportance<T>, ICloneable<IFullModel<T, Tensor<T>, Tensor<T>>>, IGradientComputable<T, Tensor<T>, Tensor<T>>, IJitCompilable<T>, IInterpretableModel<T>, IInputGradientComputable<T>, IDisposableType Parameters
TThe numeric type used for calculations (typically float or double).
- Inheritance
-
OctonionNeuralNetwork<T>
- Implements
- Inherited Members
- Extension Methods
Remarks
An Octonion Neural Network uses octonion algebra (8-dimensional non-associative division algebra) for its computations. This provides richer representational capacity than real, complex, or quaternion-valued networks, making it suitable for tasks requiring high-dimensional rotations and transformations.
For Beginners: Octonions are 8-dimensional numbers that extend complex numbers and quaternions. While regular neural networks use simple numbers, octonion networks use these 8-dimensional numbers which can capture more complex relationships in data. This is particularly useful for: - 3D graphics and physics simulations - Signal processing with multiple channels - Tasks requiring rich rotational representations
Constructors
OctonionNeuralNetwork(NeuralNetworkArchitecture<T>, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>>?, ILossFunction<T>?, double)
Initializes a new instance of the OctonionNeuralNetwork class.
public OctonionNeuralNetwork(NeuralNetworkArchitecture<T> architecture, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>>? optimizer = null, ILossFunction<T>? lossFunction = null, double maxGradNorm = 1)Parameters
architectureNeuralNetworkArchitecture<T>The architecture defining the structure of the neural network.
optimizerIGradientBasedOptimizer<T, Tensor<T>, Tensor<T>>The optimization algorithm to use for training. If null, Adam optimizer is used.
lossFunctionILossFunction<T>The loss function to use for training. If null, MSE is used.
maxGradNormdoubleThe maximum gradient norm for gradient clipping during training.
Remarks
Note: Input and output dimensions should be multiples of 8 to properly represent octonions. Each octonion has 8 real components (1 real + 7 imaginary).
Properties
SupportsTraining
Indicates whether this network supports training.
public override bool SupportsTraining { get; }Property Value
Methods
Backward(Tensor<T>)
Performs a backward pass through the network to calculate gradients.
public Tensor<T> Backward(Tensor<T> outputGradient)Parameters
outputGradientTensor<T>The gradient of the loss with respect to the network's output.
Returns
- Tensor<T>
The gradient of the loss with respect to the network's input.
CreateNewInstance()
Creates a new instance of the OctonionNeuralNetwork with the same configuration.
protected override IFullModel<T, Tensor<T>, Tensor<T>> CreateNewInstance()Returns
- IFullModel<T, Tensor<T>, Tensor<T>>
Remarks
This creates a fresh network instance with a new optimizer to avoid state conflicts. Sharing an optimizer instance between networks would cause training issues since the optimizer maintains internal state (momentum, adaptive learning rates, etc.) that is specific to each network's parameters.
DeserializeNetworkSpecificData(BinaryReader)
Deserializes octonion neural network-specific data from a binary reader.
protected override void DeserializeNetworkSpecificData(BinaryReader reader)Parameters
readerBinaryReader
Forward(Tensor<T>)
Performs a forward pass through the network with the given input tensor.
public Tensor<T> Forward(Tensor<T> input)Parameters
inputTensor<T>The input tensor to process.
Returns
- Tensor<T>
The output tensor after processing through all layers.
GetModelMetadata()
Retrieves metadata about the octonion neural network model.
public override ModelMetadata<T> GetModelMetadata()Returns
- ModelMetadata<T>
A ModelMetaData object containing information about the network.
InitializeLayers()
Initializes the layers of the octonion neural network based on the provided architecture.
protected override void InitializeLayers()IsValidInputLayer(ILayer<T>)
Determines if a layer can serve as a valid input layer for this network.
protected override bool IsValidInputLayer(ILayer<T> layer)Parameters
layerILayer<T>
Returns
IsValidOutputLayer(ILayer<T>)
Determines if a layer can serve as a valid output layer for this network.
protected override bool IsValidOutputLayer(ILayer<T> layer)Parameters
layerILayer<T>
Returns
Predict(Tensor<T>)
Makes a prediction using the octonion neural network for the given input tensor.
public override Tensor<T> Predict(Tensor<T> input)Parameters
inputTensor<T>The input tensor to make a prediction for.
Returns
- Tensor<T>
The predicted output tensor.
SerializeNetworkSpecificData(BinaryWriter)
Serializes octonion neural network-specific data to a binary writer.
protected override void SerializeNetworkSpecificData(BinaryWriter writer)Parameters
writerBinaryWriter
Train(Tensor<T>, Tensor<T>)
Trains the octonion neural network using the provided input and expected output.
public override void Train(Tensor<T> input, Tensor<T> expectedOutput)Parameters
inputTensor<T>The input tensor for training.
expectedOutputTensor<T>The expected output tensor for the given input.
UpdateParameters(Vector<T>)
Updates the parameters of all layers in the network.
public override void UpdateParameters(Vector<T> parameters)Parameters
parametersVector<T>A vector containing all parameters for the network.