Question

我目前正在尝试在python中手动实现单层感知器，并且遇到以下错误但不明白原因。

这是我的代码，以及我使用的数据集。

def predict(inVec, initWeights, bias):
    activation = bias;
    for i in range(len(inVec) - 1):
        activation += (initWeights[i] * inVec[i])
    return 1.0 if activation >= 0.0 else 0.0 

def updateWeights(inVec, weights, bias, learningRate):
    for row in inVec:
        target = row[-1]
        prediction = predict(row, weights, bias)
        error = target - prediction
        bias = bias + (learningRate*error)
        for i in range(len(row) -1):
            weights[i] = weights[i] + (learningRate*error) * row[i]
    return weights

以下是一个示例数据集：

dataset = [[10, 64.0277609352, 0], [15, 0.0383577812151, 0],[20, 22.15708796, 0],[25, 94.4005135336, 1],[30, 3.8228541672, 0],[35, 62.4202896763, 1]]

这是函数的参数：

bias = 0
weights = [0,0]
learningRate = 0.01

我希望执行以下功能：print(updateWeights(dataset, bias, weights, learningRate))

但我一直收到错误：TypeError: 'int' object is not subscriptable指向预测函数的激活更新部分。这对我来说很奇怪，因为在updateWeights函数之外单独使用该函数不会导致错误。有没有人对此有所想法？

这是完整的堆栈跟踪：

Traceback (most recent call last):
 File "/Users/x/Documents/workspace/Machine Learning/Perceptron/perceptron.py", line 53, in <module>
    print(updateWeights(dataset, bias, weights, learningRate))
  File "/Users/x/Documents/workspace/Machine Learning/Perceptron/perceptron.py", line 24, in updateWeights
    prediction = predict(row, weights, bias)
  File "/Users/x/Documents/workspace/Machine Learning/Perceptron/perceptron.py", line 17, in predict
    activation += (initWeights[i] * inVec[i])
TypeError: 'int' object is not subscriptable

实现感知器功能，获取TypeError：＆＃39; int对象不可订阅＆＃39;

0 个答案: