Machine Learning Activation Function Idea – Switch Between x And x^2 For Each Iteration

After using my imagination. I think I discovered a new class of activation functions.

def func_i(x,i): #order unknown sin tayler expansion

if i%2==0:
return x
return x**2

Just iterate between x or x**2. Like some polynom expansion.

(Visited 1 times, 1 visits today)