steepest descent method


Also found in: Acronyms.

steepest descent method

[′stēp·əst di′sent ‚meth·əd]
(mathematics)
Certain functions can be approximated for large values by an asymptotic formula derived from a Taylor series expansion about a saddle point. Also known as saddle point method.
A method of approximating extreme values of the functions of two or more variables, in which the gradient of the function is used to obtain a sequence of approximations of the point at which the extreme value occurs.
References in periodicals archive ?
So, the wiener solution for filter weight by use of steepest descent method produces: w (n + 1) = w (n) + [mu][p - Rw(n)] (9)
We used the program consisting of Steepest descent method and started the point (1,0,-1,3) as the starting point, and we reached the minimum of 3.
1 contains the steepest descent method when setting the parameter [beta] = 0.
SAMOKISH, The steepest descent method for an eigenvalue problem with semi-bounded operators, Izvestiya Vuzov, Math.
For instance, when [rho](k) = 1 for all k and v(k) = k, then the GMR reduces to the classical steepest descent method or Cauchy method.
It covers vectors, matrices, processing discrete deterministic signals: discrete systems, discrete-time random processes, the Wiener filter, eigenvalues of Rx: properties of the error surface, Newton's and steepest descent methods, the least mean-square algorithm, and variants of least mean-square algorithm.