Video details loaded
HomeMIT 18.S096 Matrix Calculus For Machine Learning And Beyond, IAP 2023Lecture 4 Part 2: Nonlinear Root Finding, Optimization, and Adjoint Gradient Methods
MIT 18.S096 Matrix Calculus For Machine Learning And Beyond, IAP 2023

Video 8 of 10

Previous: Lecture 4 Part 1: Gradients and Inner Products in Other Vector SpacesNext: Lecture 5 Part 1: Derivative of Matrix Determinant and Inverse

Lecture 4 Part 2: Nonlinear Root Finding, Optimization, and Adjoint Gradient Methods

44:26

Up Next

Lecture 5 Part 1: Derivative of Matrix Determinant and Inverse

Continue

Description: Nonlinear root finding by Newton’s method and optimization by gradient descent. “Adjoint” methods (reverse-mode/backpropagation) let us find gradients efficiently for large-scale engineering optimization.

Instructors: Alan Edelman, Steven G. Johnson