1 
# 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。关键线主要是与 SGD 的定义功能有关的那些。

4

1 回答 1

0

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，错误是因为您尝试将浮点数与列表相乘并将其添加到浮点数。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].

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

最后备注：您不需要将 EpochVersion1 转换为整数：int(EpochVersion1)。当您将其声明为 时，它已经是一个整数EpochVersion1 = 10。

于 2022-01-27T08:04:17.143 回答