斋藤康毅-深度学习入门 学习笔记三

内容导读

互联网集市收集整理的这篇技术教程文章主要介绍了斋藤康毅-深度学习入门 学习笔记三，小编现在分享给大家，供广大互联网技能从业者学习和参考。文章包含4447字，纯文字阅读大概需要7分钟。

内容图文

ch03 神经网络

1. pkl文件的创建与导入
python官方文档对pickle模块的定义：pickle

The pickle module implements binary protocols for serializing and de-serializing a Python object structure. “Pickling” is the process whereby a Python object hierarchy is converted into a byte stream, and “unpickling” is the inverse operation, whereby a byte stream (from a binary file or bytes-like object) is converted back into an object hierarchy. Pickling (and unpickling) is alternatively known as “serialization”, “marshalling,” 1 or “flattening”; however, to avoid confusion, the terms used here are “pickling” and “unpickling”.

pickle.dump(obj, file, protocol=None, *, fix_imports=True, buffer_callback=None)
Write the pickled representation of the object obj to the open file object file. This is equivalent to Pickler(file, protocol).dump(obj).

pickle.load(file, *, fix_imports=True, encoding=“ASCII”, errors=“strict”, buffers=None)
Read the pickled representation of an object from the open file object file and return the reconstituted object hierarchy specified therein. This is equivalent to Unpickler(file).load().
The protocol version of the pickle is detected automatically, so no protocol argument is needed. Bytes past the pickled representation of the object are ignored.

import _pickle as pickle

if __name__ == '__main__':
    a = {"name": "raymond",
         "age": 21}
    with open("./test.pkl", "wb") as f:
        pickle.dump(a, f)
    with open("./test.pkl", 'rb') as f:
        a = pickle.load(f)
    print(a, a.__class__)

2. 激活函数

2.1 恒等函数

def identity(x):
    return x

2.2 ReLU函数

import numpy as np
import matplotlib.pyplot as plt


def ReLU(x):
    return np.maximum(0, x)

if __name__ == '__main__':
    x = np.arange(-1.0, 1.0, 0.1)
    y = ReLU(x)
    plt.plot(x, y)
    plt.ylim(-0.1, 1.1)
    plt.show()

2.3 sigmoid函数

import numpy as np
import matplotlib.pyplot as plt


def sigmoid(x):
    return 1 / (1 + np.exp(-x))

if __name__ == '__main__':
    x = np.arange(-5.0, 5.0, 0.1)
    y = sigmoid(x)
    plt.plot(x, y)
    plt.ylim(-0.1, 1.1)
    plt.show()

2.4 softmax函数

import numpy as np


def softmax(x):
    # max 和 min效果上相同
    # 但是不使用min，因为min会导致overflow
    c = np.max(x)
    exp_x = np.exp(x - c)
    sum_exp = np.sum(exp_x)
    y = exp_x / sum_exp
    return y

# a = np.array([1, 10000, 10000])
# print(softmax(a))

2.5 阶跃函数

import numpy as np
import matplotlib.pyplot as plt


def step(x):
    return np.array(x > 0, dtype=np.int)

if __name__ == '__main__':
    x = np.arange(-5.0, 5.0, 0.1)
    y = step(x)
    plt.plot(x, y)
    plt.ylim(-0.1, 1.1)
    plt.show()

3. 神经网络前向传播的示例代码

import pickle
import sys, os
from functions.sigmoid_function import sigmoid
from functions.softmax_function import softmax

sys.path.append(os.pardir)
import numpy as np
from dataset.mnist import load_mnist


def get_data():
    (x_train, t_train), (x_test, t_test) = \
        load_mnist(normalize=True, flatten=True, one_hot_label=False)
    return x_test, t_test


# TODO:write sample_weight.pkl
def init_network():
    with open("sample_weight.pkl", "rb") as f:
        network = pickle.load(f)
    return network


def predict(network, x):
    W1, W2, W3 = network["W1"], network["W2"], network["W3"]
    b1, b2, b3 = network["b1"], network["b2"], network["b3"]
    a1 = np.dot(x, W1) + b1
    z1 = sigmoid(a1)
    a2 = np.dot(z1, W2) + b2
    z2 = sigmoid(a2)
    a3 = np.dot(z2, W3) + b3
    y = softmax(a3)
    return y


x, t = get_data()
network = init_network()
accuracy_cnt = 0
batch_size = 100
for i in range(0, len(x), batch_size):
    # 批处理加快速度
    x_batch = x[i: i + batch_size]
    y_batch = predict(network, x_batch)
    p = np.argmax(y_batch, axis=1)
    accuracy_cnt += np.sum(p == t[i: i + batch_size])
print("Accuracy:" + str(float(accuracy_cnt) / len(x)))

其中sample_weight.pkl文件可以通过如下方法生成（由于并未学习到如何通过训练得到较优的参数，所以这里是人为生成的）:

import _pickle as pickle
import numpy as np

if __name__ == '__main__':
    network = {}
    network["W1"] = np.arange(0, 784 * 50, dtype=np.int).reshape((784, 50))
    network["W2"] = np.arange(0, 50 * 100).reshape(50, 100)
    network["W3"] = np.arange(0, 100 * 10).reshape(100, 10)
    network["b1"] = np.arange(0, 50)
    network["b2"] = np.arange(0, 100)
    network["b3"] = np.arange(0, 10)

with open("../sample_weight.pkl", "wb") as f:
    pickle.dump(network, f)

总结

• 神经网络中的激活函数使用平滑的sigmoid函数或者ReLU函数
• 通过巧妙地使用Numpy多维数组，可以高效地实现神经网络
• 机器学习的问题大体上可以分为回归问题和分类问题
• 关于输出层的激活函数，回归问题一般使用恒等函数，分类问题中一般使用softmax函数
• 分类问题中，输出层的神经元的数量设置为要分类的类别数
• 输入数据的集合成为批。通过以批为单位进行推理出来，能够实现高速的运算

---

斋藤康毅-深度学习入门 专栏

内容总结

以上是互联网集市为您收集整理的斋藤康毅-深度学习入门 学习笔记三全部内容，希望文章能够帮你解决斋藤康毅-深度学习入门 学习笔记三所遇到的程序开发问题。 如果觉得互联网集市技术教程内容还不错，欢迎将互联网集市网站推荐给程序员好友。

内容备注

版权声明：本文内容由互联网用户自发贡献，该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容， 请发送邮件至 gblab@vip.qq.com 举报，一经查实，本站将立刻删除。

内容手机端

---