WaveNetOptions<T>

Options & Configuration · AiDotNet.Models.Options

Configuration options for the WaveNet model adapted for time series forecasting.

Inheritance: ModelOptions → RegressionOptions<T> → TimeSeriesRegressionOptions<T> → WaveNetOptions<T>

For Beginners

WaveNet is similar to TCN but with some key differences:

Gated Activation Units: Instead of simple ReLU activations, WaveNet uses gates:

  • tanh(Wf * x) * sigmoid(Wg * x)
  • The sigmoid acts as a “gate” controlling information flow
  • This helps model complex patterns more effectively

Skip Connections: WaveNet has TWO types of connections:

  1. Residual: Connect input to output of each block (like TCN)
  2. Skip: Each block also sends output directly to the final layers
  • This allows the network to combine features from different time scales

Stacked Dilations: WaveNet often repeats the dilation pattern multiple times:

  • [1, 2, 4, 8, 16, 1, 2, 4, 8, 16, …]
  • This creates very deep networks with huge receptive fields

Originally designed for generating audio one sample at a time, WaveNet’s architecture is now used for many sequence prediction tasks including financial forecasting.

How It Works

WaveNet was originally developed by DeepMind for audio generation but has proven highly effective for time series forecasting. It uses dilated causal convolutions with gated activations and residual/skip connections.

Reference: van den Oord et al., “WaveNet: A Generative Model for Raw Audio”, 2016. https://arxiv.org/abs/1609.03499

Constructors

Constructor Summary
WaveNetOptions() Initializes a new instance of the WaveNetOptions class with default values.
WaveNetOptions(WaveNetOptions<T> other) Initializes a new instance by copying from another instance.

Properties

Property Summary
bool UseGatedActivations { get; set; } Gets or sets whether to use gated activation units.
double DropoutRate { get; set; } Gets or sets the dropout rate for regularization.
double LearningRate { get; set; } Gets or sets the learning rate for training.
int BatchSize { get; set; } Gets or sets the batch size for training.
int DilationDepth { get; set; } Gets or sets the dilation depth (number of dilation doublings per stack).
int Epochs { get; set; } Gets or sets the number of training epochs.
int ForecastHorizon { get; set; } Gets or sets the forecast horizon (prediction length).
int KernelSize { get; set; } Gets or sets the kernel size for dilated convolutions.
int LookbackWindow { get; set; } Gets or sets the lookback window size (input sequence length).
int NumStacks { get; set; } Gets or sets the number of stacks (repetitions of the dilation pattern).
int ResidualChannels { get; set; } Gets or sets the number of residual channels.
int SkipChannels { get; set; } Gets or sets the number of skip channels.