Simultaneous perturbation stochastic approximation

Simultaneous Perturbation Stochastic Approximation (SPSA) is an optimization technique used primarily for estimating the minima or maximizing the performance of a function that is typically noisy and possibly non-differentiable. It is especially useful in situations where evaluating the function is expensive, such as in simulations, control problems, or real-world applications where measurements have inherent noise.