深度学习.

2. 基本要求 掌握深度学习的网络设计、模型训练和测试方法。

3. 支撑的课程目标 本实验项目可以支撑 “ 课程目标 2. 熟练掌握和使用结构主义的智能算法，解决复杂系统 工程的智能处理和应用问题 ” 本实验通过实验理解并体会 ANN 的设计及网络结构的学习方法，通过样本数据训练网 络，并能使用训练成熟的网络进行预测、优化等，培养学生熟悉 ANN 的典型模型、基于神 经网络的知识表示与推理 , 以及深度学习、类脑计算等方法解决实际问题的能力，达到课程 目标的要求。

4. 实验原理 MNIST 数据集下载地址： http://yann.lecun.com/exdb/mnist/ 5. 实验步骤 1) 深度学习入门（六）：手写数字识别： https://www.infoq.cn/article/fephligxt4egiokbjyet 2) 使用神经网络识别手写数字： https://hit-scir.gitbooks.io/neural-networks-and-deep-learning-zh_cn/content/chap1/c1s0 .html 3) 【深度学习系列】手写数字识别 -- 卷 积 神 经 网 络 CNN 原理详解 ( 一 ) ： https://zhuanlan.zhihu.com/p/30665319 4) matlab — — 简 单 CNN 网 络 的 数 字 识 别 : https://blog.csdn.net/xjc0518/article/details/100653760 5) python 深 度 学 习 --- 手 写 数 字 识 别 : https://blog.csdn.net/shuiyihang0981/article/details/87554765 6) Python 实现深度学习 MNIST 手写数字识别（单文件，非框架，无需 GPU ，适合初 学者） :https://blog.csdn.net/discoverer100/article/details/88047658 7) C++ 实现 CNN 识别手写数字 : https://blog.csdn.net/taoyafan/article/details/80459824 8) [ 深 度 学 习 系 列 之 一 ] C++ 实 现 神 经 网 络 : https://www.geek-share.com/detail/2716287788.html 9) C++ 实 现 一 个 识 别 MNIST 数 字 的 卷 积 神 经 网 络 : https://satanwoo.github.io/2018/05/09/MNIST/ 10) 深 度 学习 Java 类库 deeplearning4j 学习 笔记 -MNIST 手写 数字 分 类问题 : https://www.jianshu.com/p/52ffcbb2d64d 11) 【 dl4j 】 Java 深 度 学 习 入 门 之 手 写 数 字 识 别 并 生 成 exe 可 执 行 文 件 : https://www.daimajiaoliu.com/daima/4875bafcb900418

import pickle 
import gzip
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import random


def load_data():
    f = gzip.open('mnist.pkl.gz', 'rb')
    training_data, validation_data, test_data =pickle.load(f,encoding="bytes")
    f.close()
    return (training_data, validation_data, test_data)

def showimage():

    training_set, validation_set, test_set = load_data()

    flattened_images = validation_set[0]

    images = [np.reshape(f, (-1, 28)) for f in flattened_images]

    for i in range(16):
        ax = plt.subplot(4, 4, i+1)
        ax.matshow(images[i], cmap = matplotlib.cm.binary)
        plt.xticks(np.array([]))
        plt.yticks(np.array([]))
    plt.show()

class Network(object):

    def __init__(self, sizes):
        self.num_layers = len(sizes)
        self.sizes = sizes
        self.biases = [np.random.randn(y, 1) for y in sizes[1:]]
        self.weights = [np.random.randn(y, x) for x, y in zip(sizes[:-1], sizes[1:])]
        # print self.weights
        # print self.biases

    def feedforward(self, a):
        for b, w in zip(self.biases, self.weights):
            a = sigmoid(np.dot(w, a) + b)
        return a

    def SGD(self, training_data, epochs, mini_batch_size, eta, test_data=None):
        if test_data:
            test_data = list(test_data)
            n_test = len(test_data)
        training_data=list(training_data)
        n = len(training_data)
        for j in range(epochs):
            random.shuffle(training_data)
            mini_batches = [training_data[k:k + mini_batch_size]
                for k in range(0, n, mini_batch_size)]
            for mini_batch in mini_batches:
                self.update_mini_batch(mini_batch, eta)
            if test_data:
                print("Epoch {0}: {1} / {2}".format(j, self.evaluate(test_data), n_test))
            else:
                print("Epoch {0} complete".format(j))

    def update_mini_batch(self, mini_batch, eta):
        nabla_b = [np.zeros(b.shape) for b in self.biases]
        nabla_w = [np.zeros(w.shape) for w in self.weights]
        for x, y in mini_batch:
            delta_nabla_b, delta_nabla_w = self.backprop(x, y)
            nabla_b = [nb + dnb for nb, dnb in zip(nabla_b, delta_nabla_b)]
            nabla_w = [nw + dnw for nw, dnw in zip(nabla_w, delta_nabla_w)]
        self.weights = [w - (eta / len(mini_batch)) * nw
                        for w, nw in zip(self.weights, nabla_w)]
        self.biases = [b - (eta / len(mini_batch)) * nb
                       for b, nb in zip(self.biases, nabla_b)]
    def backprop(self, x, y):
        nabla_b = [np.zeros(b.shape) for b in self.biases]
        nabla_w = [np.zeros(w.shape) for w in self.weights]
        activation = x
        activations = [x]
        zs = []
        for b, w in zip(self.biases, self.weights):
            z = np.dot(w, activation) + b
            zs.append(z)
            activation = sigmoid(z)
            activations.append(activation)
        delta = self.cost_derivative(activations[-1], y) * sigmoid_prime(zs[-1])
        nabla_b[-1] = delta
        nabla_w[-1] = np.dot(delta, activations[-2].transpose())
        for l in range(2, self.num_layers):
            z = zs[-l]
            sp = sigmoid_prime(z)
            delta = np.dot(self.weights[-l + 1].transpose(), delta) * sp
            nabla_b[-l] = delta
            nabla_w[-l] = np.dot(delta, activations[-l - 1].transpose())
        return (nabla_b, nabla_w)

    def evaluate(self, test_data):
        test_results = [(np.argmax(self.feedforward(x)), y) for (x, y) in test_data]
        return sum(int(x == y) for (x, y) in test_results)

    def cost_derivative(self, output_activations, y):
        return (output_activations - y)

def sigmoid(z):
    return 1.0 / (1.0 + np.exp(-z))


def sigmoid_prime(z):
    return sigmoid(z) * (1 - sigmoid(z))


def vectorized_result(j):
    e = np.zeros((10, 1))
    e[j] = 1.0
    return e


def load_data_wrapper():
    tr_d, va_d, te_d = load_data()

    training_inputs = [np.reshape(x, (784, 1)) for x in tr_d[0]]
    training_results = [vectorized_result(y) for y in tr_d[1]]
    training_data = zip(training_inputs, training_results)
    validation_inputs = [np.reshape(x, (784, 1)) for x in va_d[0]]
    validation_data = zip(validation_inputs, va_d[1])
    test_inputs = [np.reshape(x, (784, 1)) for x in te_d[0]]
    test_data = zip(test_inputs, te_d[1])
    return (training_data, validation_data, test_data)

if __name__ == '__main__':
    training_data, valivation_data, test_data = load_data_wrapper()

    net = Network([784, 30, 10])
    net.SGD(training_data, 10, 10, 3.0, test_data=test_data)

    showimage()

实验结果