PolynomialKernel<T>
Models & Types · AiDotNet.Kernels
Implements the Polynomial kernel for measuring similarity between data points.
Implements: IKernelFunction<T>
For Beginners
A kernel function is a mathematical tool that measures how similar two data points are. The Polynomial kernel is special because it can find complex patterns in your data by implicitly mapping your data to a higher-dimensional space without actually performing the expensive calculations that would normally be required.
How It Works
The Polynomial kernel is a popular kernel function used in machine learning algorithms like Support Vector Machines (SVMs) to find patterns in higher-dimensional spaces.
Think of the Polynomial kernel as a “pattern detector” that can identify more complex relationships than simple linear patterns. For example, if you’re trying to classify data that can’t be separated by a straight line, the Polynomial kernel can help find a curved boundary instead.
The formula for the Polynomial kernel is: k(x, y) = (x·y + c)^d where:
- x and y are the two data points being compared
- x·y is the dot product (a measure of how aligned the vectors are)
- c is a constant term (coef0) that influences the kernel’s behavior
- d is the degree of the polynomial
Common uses include:
- Classification problems where data isn’t linearly separable
- Natural language processing tasks
- Image recognition
Example
using AiDotNet.Kernels;
using AiDotNet.Tensors.LinearAlgebra;
var kernel = new PolynomialKernel<double>();
var a = new Vector<double>(new[] { 1.0, 2.0, 3.0 });
var b = new Vector<double>(new[] { 1.5, 1.0, 2.5 });
double similarity = kernel.Calculate(a, b);
Console.WriteLine($"PolynomialKernel: K(a, b) = {similarity:F4}");
Constructors
| Constructor | Summary |
|---|---|
PolynomialKernel(T degree, T coef0) |
Initializes a new instance of the Polynomial kernel with optional degree and coefficient parameters. |
Methods
| Method | Summary |
|---|---|
T Calculate(Vector<T> x1, Vector<T> x2) |
Calculates the Polynomial kernel value between two vectors. |