Question

我正在尝试用Python训练感知器，该感知器具有3种不同的猫类：老虎，狮子和猎豹。为了做到这一点，我希望创建一个感知器精确度进度图。最初，我创建了3个python文件，每个文件的目的是训练每个类的感知器。下面的代码对于每个文件都是通用的-在python中，有没有一种方法可以合并这三个文件并将下面的代码实现为def？

公用代码：

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import Perceptron as nn

def commonCode(!WHAT PARAMETERS SHOULD GO HERE?!):

理想情况下，我希望在此处调用Lion和Tiger函数（见下文），但是我不确定需要调用哪些参数，也不确定如何实现此功能。

(weigths, accuracy, accuracy_progression) = nn.perceptronLearning(data,epochs,learning_rate, target_accuracy)

(tp,tn,fp,fn) = p.confusionMatrix(weigths,data)

print('weigths: ', weigths)
print('accuracy: ', accuracy)

print('true positive: %d    true negative: %d',(tp,tn))
print('false positive: %d   false negative: %d',(fp,fn))

title = "%d_iterations_lambda=%f" %(len(accuracy_progression),learning_rate)
path = "./Plots/%s.png" %(title)

plt.title(title)
plt.ylabel('accuracy (%)')
plt.xlabel('iteration')
plt.plot(accuracy_progression)
plt.show()

train_Lion.py文件：

def Lion (cat): 
    if cat == b'Cat-lion':
        return 1
    else:
        return 0

filename = 'cat.data'
data = np.loadtxt(filename,delimiter=',',converters={4:lion})
np.random.shuffle(data)

epochs = 30
learning_rate = 0.1
target_accuracy = 100

train_Tiger.py文件：

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import Perceptron as nn
def Tiger (cat): 
    if cat == b'Cat-tiger':
        return 1
    else:
        return 0

filename = 'cat.data'
data = np.loadtxt(filename,delimiter=',',converters={4:tiger})
np.random.shuffle(data)

epochs = 30
learning_rate = 0.2
target_accuracy = 95

等学习率和目标准确性在各班之间有所不同，因此我不确定是否必须将这些作为参数传递？任何建议将不胜感激！

Answer 1

我可以这样设置：

一个类，用于执行您使用纪元，learning_rate，target_accuracy和数据启动的感知器学习和绘图。然后，各个模块可以定义其特定值并实例化该类的实例。

这是您的实现的基本包装。我返回了（权重，准确性，precision_progression）元组，以便您可以根据需要在各个模块中对其进行进一步处理。您当然可以进一步重构：

import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import Perceptron as nn


class PPlotter:
    def __init__(self, epochs, learning_rate, target_accuracy, data):
        self.epochs = epochs
        self.learning_rate = learning_rate
        self.target_accuracy = target_accuracy
        self.data = data

    def plot_accuracy_progression(self):
        (weights, accuracy, accuracy_progression) = nn.perceptronLearning(self.data, self.epochs, self.learning_rate, self.target_accuracy)

        (tp,tn,fp,fn) = p.confusionMatrix(weigths,self.data)

        print('weights: ', weights)
        print('accuracy: ', accuracy)

        print('true positive: %d    true negative: %d',(tp,tn))
        print('false positive: %d   false negative: %d',(fp,fn))

        title = "%d_iterations_lambda=%f" %    (len(accuracy_progression),learning_rate)
        path = "./Plots/%s.png" %(title)
        plt.title(title)
        plt.ylabel('accuracy (%)')
        plt.xlabel('iteration')
        plt.plot(accuracy_progression)
        plt.show()

        return (weights, accuracy, accuracy_progression)

然后是一个train_Lion.py外观的示例：

import numpy as np
import PPlotter

def Lion (cat):
    if cat == b'Cat-lion':
        return 1
    else:
        return 0

filename = 'cat.data'
data = np.loadtxt(filename,delimiter=',',converters={4:lion})
np.random.shuffle(data)

epochs = 30
learning_rate = 0.1
target_accuracy = 100

plotter = PPlotter(epochs, learning_rate, target_accuracy, data)

(weights, accuracy, accuracy_progression) = plotter.plot_accuracy_progression()

在另一个Python中调用一个def

1 个答案: