steepest descent method; gradient descent method; saddle point method
Mathematical optimization algorithm that finds a local minimum by moving in the direction of the negative gradient. In Japanese, 最急降下法 is the standard term; 'saddle point method' is a less common English equivalent sometimes used in physics contexts.
The steepest descent method is an optimization algorithm often used in machine learning.
We applied the steepest descent method to find the minimum of this function.
勾配降下法 is a synonym for 最急降下法, both meaning gradient descent. 最急降下法 emphasizes the 'steepest' direction, while 勾配降下法 is more literally 'gradient descent'.
Newton's method uses second-order derivative information (Hessian) for faster convergence, while 最急降下法 uses only first-order gradients.
Compound of 最急 (steepest), 降下 (descent), and 法 (method). A direct translation of the English mathematical term 'steepest descent method'.