[Retracted] Development of Basketball Tactics Basic Cooperation Teaching System Based on CNN and BP Neural Network

Figure 10

Training effect of various training algorithms. (a) Weight adjustment times. (b) Mean square error value. (c) Error surface gradient. (d) Recognition rate. I, II, III, IV, and V refer to gradient descent method, adaptive learning rate gradient descent method, driving amount gradient descent method, Powell Beale conjugate gradient method, and quasi-Newton method, respectively.

(a)

(b)

(c)

(d)