Encountering a TypeError: can't multiply sequence by non-int of type 'float' when creating an SGD algorithm

Question

# We first define the observations as a list and then also as a table for the experienced worker's performance.

Observation1 = [2.0, 6.0, 2.0]
Observation2 = [1.0, 5.0, 7.0]
Observation3 = [5.0, 2.0, 1.0]
Observation4 = [2.0, 3.0, 8.0]
Observation5 = [4.0, 4.0, 0.0]
ObservationTable = [
                    Observation1,
                    Observation2,
                    Observation3,
                    Observation4,
                    Observation5
                    ]

# Then we define our learning rate, number of observations, and the epoch counters we will be utilizing (10, 100, and 1000).

LearningRate = 0.01
ObservationCounter = 5
EpochVersion1 = 10
EpochVersion2 = 100
EpochVersion3 = 1000

# Thus, we are now ready to define the Stochastic Gradient Descent Algorithm:

def StochasticGradientDescent(EpochCounter):
    Theta0 = 10.0
    Theta1 = 0.0
    Theta2 = -1.0

    while (EpochCounter != 0):

        ObservationCounter = 5

        while (ObservationCounter >= 0):

            Theta0_Old = float(Theta0)
            Theta1_Old = float(Theta1)
            Theta2_Old = float(Theta2)

            n = 5 - ObservationCounter
            x = ObservationTable [n]
            x0 = float(x[0])
            x1 = float(x[1])
            x2 = float(x[2])

            Theta0_New = Theta0_Old - LearningRate*[(Theta0_Old+Theta1_Old*float(x0)+Theta2_Old*float(x1))-float(x2)]
            Theta1_New = Theta1_Old - LearningRate*[(Theta0_Old+Theta1_Old*float(x0)+Theta2_Old*float(x1))-float(x2)]*float(x0)
            Theta2_New = Theta2_Old - LearningRate*[(Theta0_Old+Theta1_Old*float(x0)+Theta2_Old*float(x1))-float(x2)]*float(x1)

            print(Theta0_New, Theta1_New, Theta2_New)
            ObservationCounter -= 1

        else:
            EpochCounter -= 1
            if (EpochCounter == 0):
               print(Theta0_New, Theta1_New, Theta2_New)


StochasticGradientDescent(int(EpochVersion1))

代码输出 TypeError: can't multiply sequence by non-int of type 'float'。我已经在每个可能的步骤将值转换为浮点数，但错误仍然存在。关键行主要是与 SGD 的定义函数有关的行。

Answer 1

我不太了解随机梯度下降，但我在您的代码中发现了两个改进。

首先，错误是因为你试图将一个浮点数与一个列表相乘并将其添加到一个浮点数中。这是通过使用圆括号而不是方括号来解决的：

            Theta0_New = Theta0_Old - LearningRate*((Theta0_Old+Theta1_Old*float(x0)+Theta2_Old*float(x1))-float(x2))
            Theta1_New = Theta1_Old - LearningRate*((Theta0_Old+Theta1_Old*float(x0)+Theta2_Old*float(x1))-float(x2))*float(x0)
            Theta2_New = Theta2_Old - LearningRate*((Theta0_Old+Theta1_Old*float(x0)+Theta2_Old*float(x1))-float(x2))*float(x1)

其次，您的 while 循环应该提前结束一次迭代，否则当您尝试访问 ObservationTable[5].

时会出错

因此将您的 while 循环更改为：

while (ObservationCounter >= 1):

输出：

9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.98 -0.04 -1.12
10.02 0.02 -0.9
9.93 -0.35000000000000003 -1.1400000000000001
10.01 0.02 -0.97
9.94 -0.24 -1.24
9.94 -0.24 -1.24

最后的评论：您不需要将 EpochVersion1 转换为整数：int(EpochVersion1)。当您将其声明为 EpochVersion1 = 10.

时，它已经是一个整数

Encountering a TypeError: can't multiply sequence by non-int of type 'float' when creating an SGD algorithm

Encountering a TypeError: can't multiply sequence by non-int of type 'float' when creating an SGD algorithm

python

floating-point

typeerror

while-loop

stochastic-gradient