Automatic Differentiation of Algorithms for Machine Learning

TL;DR

This paper reviews automatic differentiation, a technique for efficiently computing derivatives, highlighting its two main modes and potential benefits for machine learning practitioners, aiming to make it accessible to a broad audience.

Contribution

It provides a comprehensive overview of automatic differentiation, emphasizing its relevance and application in machine learning, and clarifies its two main modes for practitioners.

Findings

Automatic differentiation predates and generalizes backpropagation.

Practitioners underutilize automatic differentiation tools.

The paper makes the technique accessible to non-experts.

Abstract

Automatic differentiation---the mechanical transformation of numeric computer programs to calculate derivatives efficiently and accurately---dates to the origin of the computer age. Reverse mode automatic differentiation both antedates and generalizes the method of backwards propagation of errors used in machine learning. Despite this, practitioners in a variety of fields, including machine learning, have been little influenced by automatic differentiation, and make scant use of available tools. Here we review the technique of automatic differentiation, describe its two main modes, and explain how it can benefit machine learning practitioners. To reach the widest possible audience our treatment assumes only elementary differential calculus, and does not assume any knowledge of linear algebra.