CoordinateDescentOptimizerOptions<T, TInput, TOutput>

Options & Configuration · AiDotNet.Models.Options

Configuration options for the Coordinate Descent optimization algorithm, which optimizes a function by solving for one variable at a time while holding others constant.

Inheritance: ModelOptions → OptimizationAlgorithmOptions<T, TInput, TOutput> → GradientBasedOptimizerOptions<T, TInput, TOutput> → CoordinateDescentOptimizerOptions<T, TInput, TOutput>

For Beginners

Imagine you’re trying to find the lowest point in a valley. Coordinate Descent is like first walking only north/south until you can’t go any lower, then switching to only east/west, then back to north/south, and so on. By taking turns moving in different directions, you can eventually reach the bottom of the valley. This approach can be simpler than trying to move in all directions at once.

How It Works

Coordinate Descent is an optimization technique that minimizes a function by updating one coordinate (or variable) at a time, while keeping all other coordinates fixed. This approach can be effective for problems where optimizing along individual dimensions is easier than optimizing all dimensions simultaneously.

Constructors

Constructor Summary
CoordinateDescentOptimizerOptions()

Properties

Property Summary
double LearningRateDecreaseRate { get; set; } Gets or sets the rate at which the learning rate decreases when performance worsens.
double LearningRateIncreaseRate { get; set; } Gets or sets the rate at which the learning rate increases when performance improves.
double MomentumDecreaseRate { get; set; } Gets or sets the rate at which momentum decreases when performance worsens.
double MomentumIncreaseRate { get; set; } Gets or sets the rate at which momentum increases when performance improves.
int BatchSize { get; set; } Gets or sets the batch size for mini-batch gradient estimation.