StudentTGaussianProcess<T>
Models & Types · AiDotNet.GaussianProcesses
Implements a Gaussian Process with Student-t likelihood for robust regression.
Inheritance: ModelBase<T, Matrix<T>, Vector<T>> → GaussianProcessBase<T> → StudentTGaussianProcess<T>
Implements: IGeneratedParameterRegistrar<T>
For Beginners
Standard Gaussian Processes assume Gaussian (normal) noise, which is sensitive to outliers. A single outlier can drastically affect predictions.
The Student-t GP uses a heavy-tailed Student-t distribution instead:
- More tolerant of outliers (they have less influence)
- The “degrees of freedom” parameter (ν) controls robustness:
- ν = 1: Cauchy distribution (very robust, heavy tails)
- ν = 4-5: Good balance of robustness and efficiency
- ν → ∞: Approaches Gaussian (standard GP)
This uses Expectation Propagation (EP) for approximate inference since the Student-t likelihood makes exact inference intractable.
How It Works
When to use:
- Data with potential outliers
- Sensor data with occasional erroneous readings
- Financial data with market anomalies
- Any regression where robustness to bad data points is important
Constructors
| Constructor | Summary |
|---|---|
StudentTGaussianProcess(IKernelFunction<T> kernel, double nu, double scale, int maxIterations, double tolerance, double damping, double jitter) |
Initializes a new Student-t Gaussian Process. |
Properties
| Property | Summary |
|---|---|
IKernelFunction<T> Kernel { get; } |
Gets the kernel function. |
double DegreesOfFreedom { get; } |
Gets the degrees of freedom. |
double Scale { get; } |
Gets the noise scale parameter. |
Methods
| Method | Summary |
|---|---|
ValueTuple<T, T> Predict(Vector<T> x) |
Predicts the mean and variance for a single input point. |
ValueTuple<Vector<T>, Vector<T>> PredictBatch(Matrix<T> XNew) |
Makes predictions with uncertainty estimates for multiple points. |
Vector<T> GetOutlierWeights() |
Gets the outlier weights after training. |
int[] GetOutlierIndices(double threshold) |
Identifies outliers in the training data. |
void Fit(Matrix<T> X, Vector<T> y) |
Trains the Student-t GP using Expectation Propagation. |
void UpdateKernel(IKernelFunction<T> kernel) |
Updates the kernel function. |