多层神经网络实现鸢尾花分类

IT小辉同学

已于 2023-04-27 10:47:41 修改

阅读量1.4k

点赞数 4

分类专栏：基础扎实文章标签：神经网络分类深度学习

于 2022-10-04 22:04:32 首次发布

本文链接：https://blog.csdn.net/weixin_53742691/article/details/127167119

版权

基础扎实专栏收录该内容

120 篇文章 3 订阅

订阅专栏

多层神经网络实现鸢尾花分类

所有文章不设限，我们相遇偶然，相散坦然，互不打扰，各自安好，向阳而生

导入鸢尾花数据

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

dataset = pd.read_csv("./data/datalist/iris.csv")

对于数据集进行观察

import seaborn as sns

sns.pairplot(dataset.iloc[:, 1:6], hue="Species")

在这里插入图片描述

生成准备数据集

X = dataset.iloc[:, 1:5].values
y = dataset.iloc[:, 5].values

数据集处理

from sklearn.preprocessing import LabelEncoder

encoder = LabelEncoder()
y1 = encoder.fit_transform(y)
print("处理后的数据集\n", y1[:5])
# 将y1转换为神经网络需要使用的数组结构
Y = pd.get_dummies(y1).values
print("处理后的数据集\n", Y[:5])

处理后的数据集
 [0 0 0 0 0]
处理后的数据集
 [[1 0 0]
 [1 0 0]
 [1 0 0]
 [1 0 0]
 [1 0 0]]

数据集分割

from sklearn.model_selection import train_test_split

X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2, random_state=0)

创建神经网络模型

导入函数库

from keras.layers import Dense
from keras.models import Sequential
from tensorflow.keras.optimizers import Adam

设置网络模型

model = Sequential()
# 输入层
model.add(Dense(4, input_shape=(4,), activation="relu"))
# 第一层（8个神经元）
model.add(Dense(8, activation="relu"))
# 第二层（6个神经元）
model.add(Dense(6, activation="relu"))
# 输出层
model.add(Dense(3, activation="softmax"))

设置网络参数

model.compile(
    # 设置学习率
    Adam(learning_rate=0.04),
    # 设置损失函数
    "categorical_crossentropy",
    # 指定准确率计算方式
    metrics=["accuracy"],
)

查看网络结构

model.summary()

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense (Dense)                (None, 4)                 20        
_________________________________________________________________
dense_1 (Dense)              (None, 8)                 40        
_________________________________________________________________
dense_2 (Dense)              (None, 6)                 54        
_________________________________________________________________
dense_3 (Dense)              (None, 3)                 21        
=================================================================
Total params: 135
Trainable params: 135
Non-trainable params: 0
_________________________________________________________________

模型训练

history = model.fit(X_train, Y_train, epochs=500)

Epoch 1/500
4/4 [==============================] - 1s 5ms/step - loss: 1.4580 - accuracy: 0.3833
Epoch 2/500
4/4 [==============================] - 0s 5ms/step - loss: 1.0303 - accuracy: 0.2000
Epoch 3/500
4/4 [==============================] - 0s 5ms/step - loss: 0.9005 - accuracy: 0.5083
Epoch 4/500
4/4 [==============================] - 0s 4ms/step - loss: 0.7248 - accuracy: 0.6917
Epoch 5/500
4/4 [==============================] - 0s 6ms/step - loss: 0.5244 - accuracy: 0.7333
Epoch 6/500
4/4 [==============================] - 0s 6ms/step - loss: 0.3771 - accuracy: 0.8750
Epoch 7/500
4/4 [==============================] - 0s 5ms/step - loss: 0.2725 - accuracy: 0.9583
Epoch 8/500
4/4 [==============================] - 0s 6ms/step - loss: 0.2024 - accuracy: 0.9583
Epoch 9/500
4/4 [==============================] - 0s 5ms/step - loss: 0.2492 - accuracy: 0.8917
Epoch 10/500
4/4 [==============================] - 0s 5ms/step - loss: 0.1896 - accuracy: 0.9250
Epoch 11/500
4/4 [==============================] - 0s 6ms/step - loss: 0.1430 - accuracy: 0.9417
Epoch 12/500
4/4 [==============================] - 0s 5ms/step - loss: 0.1202 - accuracy: 0.9500
Epoch 13/500
4/4 [==============================] - 0s 5ms/step - loss: 0.1458 - accuracy: 0.9417
Epoch 14/500
4/4 [==============================] - 0s 4ms/step - loss: 0.1651 - accuracy: 0.9417
Epoch 15/500
4/4 [==============================] - 0s 5ms/step - loss: 0.1096 - accuracy: 0.9583
Epoch 16/500
4/4 [==============================] - 0s 4ms/step - loss: 0.1139 - accuracy: 0.9583
Epoch 17/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0925 - accuracy: 0.9667
Epoch 18/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0837 - accuracy: 0.9583
Epoch 19/500
4/4 [==============================] - 0s 12ms/step - loss: 0.0887 - accuracy: 0.9667
Epoch 20/500
4/4 [==============================] - 0s 5ms/step - loss: 0.1162 - accuracy: 0.9583
Epoch 21/500
4/4 [==============================] - 0s 4ms/step - loss: 0.1475 - accuracy: 0.9333
Epoch 22/500
4/4 [==============================] - 0s 4ms/step - loss: 0.2753 - accuracy: 0.9250
Epoch 23/500
4/4 [==============================] - 0s 5ms/step - loss: 0.1395 - accuracy: 0.9250
Epoch 24/500
4/4 [==============================] - 0s 4ms/step - loss: 0.1020 - accuracy: 0.9583
Epoch 25/500
4/4 [==============================] - 0s 4ms/step - loss: 0.1085 - accuracy: 0.9583
Epoch 26/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0826 - accuracy: 0.9667
Epoch 27/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0724 - accuracy: 0.9750
Epoch 28/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0723 - accuracy: 0.9750
Epoch 29/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0818 - accuracy: 0.9667
Epoch 30/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0802 - accuracy: 0.9750
Epoch 31/500
4/4 [==============================] - 0s 4ms/step - loss: 0.1195 - accuracy: 0.9583
Epoch 32/500
4/4 [==============================] - 0s 5ms/step - loss: 0.0833 - accuracy: 0.9667
Epoch 33/500
4/4 [==============================] - 0s 4ms/step - loss: 0.0998 - accuracy: 0.9583
Epoch 34/500

…

4/4 [==============================] - 0s 5ms/step - loss: 0.0693 - accuracy: 0.9750
Epoch 499/500
4/4 [==============================] - 0s 5ms/step - loss: 0.0524 - accuracy: 0.9750
Epoch 500/500
4/4 [==============================] - 0s 11ms/step - loss: 0.0916 - accuracy: 0.9500

训练结果可视化

hist = pd.DataFrame(history.history)
hist.head()

	loss	accuracy
0	1.457973	0.383333
1	1.030313	0.200000
2	0.900486	0.508333
3	0.724796	0.691667
4	0.524401	0.733333

绘制损失值变化趋势

import matplotlib.pyplot as plt

plt.figure(figsize=(6, 3))
plt.plot(hist.index, hist["loss"])
plt.show()

在这里插入图片描述

绘制准确率变化趋势

import matplotlib.pyplot as plt

plt.figure(figsize=(6, 3))
plt.plot(hist.index, hist["accuracy"])
plt.show()

在这里插入图片描述

模型预测

y_pred = model.predict(X_test)

预测值处理（获取最大概率index）

y_pred_class = np.argmax(y_pred, axis=1)

Y_test_class = np.argmax(Y_test, axis=1)

模型评估

count = 0
for i in range(len(y_pred_class)):
    if y_pred_class[i] == Y_test_class[i]:
        count = count + 1
print("正确率：", count / len(Y_test))

正确率： 1.0

多层神经网络实现鸢尾花分类2

import matplotlib as mpl
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import tensorflow as tf

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib as mpl
import tensorflow as tf

# 1，数据获取
# 加载训练集
TRAIN_URL="http://download.tensorflow.org/data/iris_training.csv"
train_path=tf.keras.utils.get_file(TRAIN_URL.split('/')[-1],TRAIN_URL)
# 加载测试集
TEST_URL="http://download.tensorflow.org/data/iris_test.csv"
test_path=tf.keras.utils.get_file(TEST_URL.split('/')[-1],TEST_URL)

# 使用pandas读取数据，得到FataFram格式数据
df_iris_train=pd.read_csv(train_path,header=0)
df_iris_test=pd.read_csv(test_path,header=0)
# 转换成numpy格式
iris_train=np.array(df_iris_train)
iris_test=np.array(df_iris_test)

# 2，数据预处理
# 2.1拆分样本特征和标签
x_train = iris_train[:,0:4]
y_train = iris_train[:,4]
# print(x_train.shape,y_train.shape)

x_test = iris_test[:,0:4]
y_test = iris_test[:,4]
# print(x_test.shape,y_test.shape)

# 2.2数据归一化
# 由于样本的4个征征值尺度相同，不用进行归一化。

# 2.3数据中心化
# 需要按列中心化，所以指定axis=0
x_train=x_train-np.mean(x_train,axis=0)
x_test=x_test-np.mean(x_test,axis=0)
# print(x_train)
# print(x_test)

# 2.4类型转换
# 将训练集特征转为float32类型
X_train=tf.cast(x_train,tf.float32)
# 将训练集标签转为int32，再转为独热编码
Y_train=tf.one_hot(tf.constant(y_train,tf.int32),3)
# print(Y_train[0:4,:])
# 将测试特征转为float32类型
X_test=tf.cast(x_test,tf.float32)
# 将测试标签转为int32，再转为独热编码
Y_test=tf.one_hot(tf.constant(y_test,tf.int32),3)

# 3，设置超参数
# 学习率
learn_rate=0.5
# 迭代次数
iter=50
# 显示频率
display_step=10
# 模型参数
np.random.seed(612)
W1=tf.Variable(np.random.randn(4,16),dtype=tf.float32)
B1=tf.Variable(np.zeros([16]),dtype=tf.float32)
W2=tf.Variable(np.random.randn(16,3),dtype=tf.float32)
B2=tf.Variable(np.zeros([3]),dtype=tf.float32)

# 4，训练模型
# 训练准确率
acc_train=[]
# 测试准确率
acc_test=[]
# 训练损失
cce_train=[]
# 测试损失
cce_test=[]

for i in range(0,iter+1):
    with tf.GradientTape() as tape:
        # 训练集隐含层线性结果
        hidden_train = tf.matmul(X_train,W1)+B1
        # 训练集隐含层输出
        Hidden_train=tf.nn.relu(hidden_train)
        # 训练集输出层线性结果
        pred_train = tf.matmul(Hidden_train,W2)+B2
        # 训练集输出层输出
        PRED_train=tf.nn.softmax(pred_train)
        # 训练集交叉熵损失
        Loss_train=tf.reduce_mean(tf.keras.metrics.categorical_crossentropy(y_true=Y_train,y_pred=PRED_train))
        
    # 测试集隐含层线性输出
    Hidden_test=tf.nn.relu(tf.matmul(X_test,W1)+B1)
    # 测试集概率输出
    PRED_test=tf.nn.softmax(tf.matmul(Hidden_test,W2)+B2)
    # 测试集交叉熵损失
    Loss_test=tf.reduce_mean(tf.keras.metrics.categorical_crossentropy(y_true=Y_test,y_pred=PRED_test))
    
    # 添加交叉熵损失到列表
    cce_train.append(Loss_train)
    cce_test.append(Loss_test)
    # 计算准确率
    accuracy_train=tf.reduce_mean(tf.cast(tf.equal(tf.argmax(PRED_train.numpy(),1),y_train),tf.float32))
    accuracy_test=tf.reduce_mean(tf.cast(tf.equal(tf.argmax(PRED_test.numpy(),1),y_test),tf.float32))
    # 添加准确率到数组
    acc_train.append(accuracy_train)
    acc_test.append(accuracy_test)
    # 计算偏导数
    grads = tape.gradient(Loss_train,[W1,B1,W2,B2])
    # 更新参数
    W1.assign_sub(learn_rate*grads[0])
    B1.assign_sub(learn_rate*grads[1])
    W2.assign_sub(learn_rate*grads[2])
    B2.assign_sub(learn_rate*grads[3])
    if i % display_step==0:
        print(i,',训练集准确率:',accuracy_train.numpy(),',训练集损失:',Loss_train.numpy(),',测试集准确率:',accuracy_test.numpy(),',测试集损失:',Loss_test.numpy())
        
# 5,可视化结果
plt.figure(figsize=(10,3))
# 5.1绘制损失
plt.subplot(121)
# 绘制训练集损失
plt.plot(cce_train,color="blue",label="train")
# 绘制测试集损失
plt.plot(cce_test,color="red",label="test")
plt.xlabel("Iter")
plt.ylabel("cce")
plt.legend(["train","test"])

# 5.2绘制准确率
plt.subplot(122)
# 绘制训练集准确率
plt.plot(acc_train,color="blue",label="train")
# 绘制测试集准确率
plt.plot(acc_test,color="red",label="test")
plt.xlabel("Iter")
plt.ylabel("acc")
plt.legend(["train","test"])
plt.show()

0 ,训练集准确率: 0.43333334 ,训练集损失: 2.2056413 ,测试集准确率: 0.4 ,测试集损失: 1.721138
10 ,训练集准确率: 0.94166666 ,训练集损失: 0.20531389 ,测试集准确率: 0.96666664 ,测试集损失: 0.24966088
20 ,训练集准确率: 0.95 ,训练集损失: 0.14953992 ,测试集准确率: 1.0 ,测试集损失: 0.16710347
30 ,训练集准确率: 0.9583333 ,训练集损失: 0.12234607 ,测试集准确率: 1.0 ,测试集损失: 0.12469266
40 ,训练集准确率: 0.9583333 ,训练集损失: 0.10509937 ,测试集准确率: 1.0 ,测试集损失: 0.09986884
50 ,训练集准确率: 0.9583333 ,训练集损失: 0.092933945 ,测试集准确率: 1.0 ,测试集损失: 0.08488502

在这里插入图片描述