Python || Local Minima

Local minima: 指的是在做梯度下降時，最佳點落在了局部最低處，而非 Global Minima。可以用 Momentum 來降低困在 Local Minima 的機率。

def FindLocalMinima(numbers):

  minima = []
  length = len(numbers)
  if length >= 2:
    if numbers[0] < numbers[1]:
      minima.append(numbers.index(numbers[0]))
       
    if length > 3:
      for i in range(1, length-1):     
        if numbers[i] < numbers[i-1] and numbers[i] < numbers[i+1]:
          minima.append(numbers.index(numbers[i]))

    if numbers[length-1] < numbers[length-2]:    
      minima.append(numbers.index(numbers[length-1]))
     
  return minima
   
   
numbers = [2, 7, 45, 3, 6, 98, 12, 67, 8]
FindLocalMinima(numbers)

[0, 3, 6, 8]

如果我們假定非字串而是隨機整數去尋找local minima

def FindLocalMinima(numbers):

  minima = []
  length = len(numbers)
  if length >= 2:
    if numbers[0] < numbers[1]:
      minima.append(numbers[0])
       
    if length > 3:
      for i in range(1, length-1):     
        if numbers[i] < numbers[i-1] and numbers[i] < numbers[i+1]:
          minima.append(numbers[i])

    if numbers[length-1] < numbers[length-2]:    
      minima.append(numbers[length-
     
  return minima
   
rand_ary = rd.sample(range(1,2000000),1000000)
print("million local minima:")
print(FindLocalMinima(rand_ary))