gradient ascent

Terms from Artificial Intelligence: humans at the heart of algorithms

The glossary is being gradually proof checked, but currently has many typos and misspellings.

Gradient ascent tries to find the highest point in a fitness landscape by following the direction with the fastest gain in value. It is equivalent to gradient descent, but with the fitness function inverted. Hill climbing is the simplest use of gradient ascent. Some algorithms may use gradient ascent mixed with other methods.

Used in Chap. 9: page 122