ConvexSolverType

Enums · AiDotNet.Enums

Types of convex solvers available for MetaOptNet.

For Beginners

Instead of iteratively updating weights, convex solvers find the optimal solution directly. Think of it like finding the lowest point in a bowl - convex problems have a single lowest point, so we can find it mathematically without searching around.

How It Works

MetaOptNet uses convex optimization in the inner loop instead of gradient descent. These solver types provide different trade-offs between speed and classification power.

Fields

Field Summary
ConvexSolverType LogisticRegression Logistic regression solved via Newton’s method.
ConvexSolverType RidgeRegression Ridge regression (L2-regularized least squares).
ConvexSolverType SVM Support Vector Machine with quadratic programming.
int value__