MLlib Module Code Review

MLlib Module Code Review

Outline

Framework

Gradient

Class used to compute the gradient for a loss function, given a single data point.

Updater

Class used to perform steps (weight update) using Gradient Descent methods.

Optimizer

Trait for optimization problem solvers.

GradientDescent(SGD)

Class used to solve an optimization problem using Gradient Descent.

L-BFGS (limited-memory BFGS)

Class used to solve an optimization problem using Limited-memory BFGS.

NNLS

Object used to solve nonnegative least squares problems using a modified projected gradient method.

Reference

Spark Programming Guide About MLlib - Optimization