AI：基于TensorFlow的Keras训练神经网络模型三分分类预测，Python

小鱼儿 2023-06-13 03:29 1阅读 0赞

# AI：基于TensorFlow的Keras训练神经网络模型三分分类预测，Python #

坐标体系内一批随机数据样本(x,y)点，实现一个简单功能。

基本分类规则：

（1）每个接近坐标原点(0,0)的坐标样本点分类标签为1。

（2）那些远离坐标原点(0,0)，而接近坐标点(1,1)的样本分类标签为0。

然后随机生成若干样本点，让借助神经网络训练后的模型进行预测，输出分类结果。

Python代码：

from keras.models import Sequential
    from keras.layers import Dense
    from sklearn.datasets.samples_generator import make_blobs
    from sklearn.preprocessing import MinMaxScaler
    from matplotlib import pyplot
    
    
    def printer(x, y):
        for i in range(len(x)):
            print("x=%s 预测为:%s" % (x[i], y[i]))
    
    
    CLASS = 2
    
    #制造一批预训练数据。
    x, y = make_blobs(n_samples=100, centers=CLASS, n_features=2, random_state=1)
    
    scaler = MinMaxScaler()
    # MinMaxScaler：归一到 [ 0，1 ]
    scaler.fit(x)
    x = scaler.transform(x)
    
    print("原始训练数据")
    for i in range(len(x)):
        print("x=%s 标签:%s" % (x[i], y[i]))
    print("-----")
    
    pyplot.scatter(x[:, 0], x[:, 1], c=y)
    
    # 构建神经网络
    model = Sequential()
    model.add(Dense(units=4, input_dim=2, activation="relu"))
    model.add(Dense(units=4, activation="relu"))
    model.add(Dense(units=1, activation="sigmoid"))
    
    # 编译模型
    model.compile(loss="binary_crossentropy", optimizer="sgd")
    
    # 训练模型
    model.fit(x, y, epochs=500, verbose=0)
    
    # n_samples条新数据，测试模型训练结果如何
    x_new, _ = make_blobs(n_samples=2, centers=CLASS, n_features=2, random_state=1)
    x_new = scaler.transform(x_new)
    y_new = model.predict_classes(x_new)
    
    print("*****")
    # 打印测试数据点
    printer(x_new, y_new)
    
    # 用红色的点画出测试数据的点。
    pyplot.scatter(x_new[:, 0], x_new[:, 1], c='r')
    
    # 用文本画出测试数据的分类。
    for i in range(CLASS):
        pyplot.text(x_new[:, 0][i], x_new[:, 1][i], s=str(y_new[i]), size=22, color='g')
    
    pyplot.show()

程序运行后，输出的图：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly96aGFuZ3BoaWwuYmxvZy5jc2RuLm5ldA_size_16_color_FFFFFF_t_70][]

输出的文本：

原始训练数据
    x=[0.89337759 0.65864154] 标签:0
    x=[0.21839815 0.12216597] 标签:1
    x=[0.70685073 0.79800428] 标签:0
    x=[0.84206699 0.89621938] 标签:0
    x=[0.35425701 0.16953972] 标签:1
    x=[0.21482032 0.19243259] 标签:1
    x=[0.20498394 0.16402074] 标签:1
    x=[0.21408907 0.14046659] 标签:1
    x=[0.73459283 0.92589868] 标签:0
    x=[0.86293838 0.83126495] 标签:0
    x=[0.14693337 0.0635184 ] 标签:1
    x=[0.83332236 0.92471684] 标签:0
    x=[0.14113841 0.14094497] 标签:1
    x=[0.76313985 0.96837513] 标签:0
    x=[0.32698889 0.0417588 ] 标签:1
    x=[0.22971925 0.20538083] 标签:1
    x=[0.16883584 0.24175307] 标签:1
    x=[0.76798454 0.77157996] 标签:0
    x=[0.79571246 0.84050646] 标签:0
    x=[0.06162881 0.20195689] 标签:1
    x=[0.19550959 0.20749043] 标签:1
    x=[0.12855764 0.19135056] 标签:1
    x=[0.17015267 0.16581683] 标签:1
    x=[0.84925052 0.81778689] 标签:0
    x=[0.79317393 0.78759208] 标签:0
    x=[0.12501222 0.        ] 标签:1
    x=[0.18570507 0.18105428] 标签:1
    x=[0.8477266 0.8098086] 标签:0
    x=[0.94156966 0.67736192] 标签:0
    x=[0.19210256 0.1000778 ] 标签:1
    x=[0.80184776 0.87825451] 标签:0
    x=[0.73120888 0.81003405] 标签:0
    x=[0.10829582 0.11150202] 标签:1
    x=[0.80799009 0.76829003] 标签:0
    x=[0.28151464 0.21503406] 标签:1
    x=[0.13499041 0.18093093] 标签:1
    x=[0.92031331 0.85148438] 标签:0
    x=[0.03915974 0.21326187] 标签:1
    x=[0.13544669 0.14198735] 标签:1
    x=[0.13903872 0.17958867] 标签:1
    x=[0.1727263  0.18315143] 标签:1
    x=[0.17837018 0.08705392] 标签:1
    x=[0.79744376 0.80734217] 标签:0
    x=[0.05071381 0.23151541] 标签:1
    x=[0.83949909 0.84635463] 标签:0
    x=[0.19282048 0.19950001] 标签:1
    x=[0.89629989 0.87609389] 标签:0
    x=[1.         0.72882744] 标签:0
    x=[0.73324478 0.85530443] 标签:0
    x=[0.81355841 0.76455675] 标签:0
    x=[0.80552738 0.82155677] 标签:0
    x=[0.20328627 0.0626137 ] 标签:1
    x=[0.15634375 0.2367766 ] 标签:1
    x=[0.11768194 0.25272282] 标签:1
    x=[0.83318811 0.88498954] 标签:0
    x=[0.19005398 0.1752731 ] 标签:1
    x=[0.77340963 0.89125781] 标签:0
    x=[0.25293258 0.34136975] 标签:1
    x=[0.82689292 0.88232276] 标签:0
    x=[0.86490997 0.81400847] 标签:0
    x=[0.83855105 0.86892497] 标签:0
    x=[0.09781082 0.25873148] 标签:1
    x=[0.19069795 0.28911078] 标签:1
    x=[0.14807041 0.12870764] 标签:1
    x=[0.         0.28468564] 标签:1
    x=[0.25759277 0.16654062] 标签:1
    x=[0.76255514 0.93428527] 标签:0
    x=[0.04728128 0.21313688] 标签:1
    x=[0.80955751 0.76904774] 标签:0
    x=[0.85768678 0.84312324] 标签:0
    x=[0.1826463  0.23485323] 标签:1
    x=[0.2098691  0.11486197] 标签:1
    x=[0.19870447 0.22141413] 标签:1
    x=[0.89124345 0.90933235] 标签:0
    x=[0.9575363  0.89467094] 标签:0
    x=[0.84897559 0.68030412] 标签:0
    x=[0.22023356 0.24701562] 标签:1
    x=[0.15358548 0.17127067] 标签:1
    x=[0.78082614 0.75391697] 标签:0
    x=[0.78737153 0.93208592] 标签:0
    x=[0.9148788  0.95498709] 标签:0
    x=[0.05320055 0.25577361] 标签:1
    x=[0.83640884 0.9050723 ] 标签:0
    x=[0.30388699 0.27498359] 标签:1
    x=[0.76762207 0.80635735] 标签:0
    x=[0.91505164 0.75182912] 标签:0
    x=[0.83890271 1.        ] 标签:0
    x=[0.89624055 0.78410253] 标签:0
    x=[0.79875321 0.90133484] 标签:0
    x=[0.76927084 0.83614339] 标签:0
    x=[0.24377251 0.21747541] 标签:1
    x=[0.82758701 0.78772661] 标签:0
    x=[0.80661236 0.88261807] 标签:0
    x=[0.76931799 0.86640554] 标签:0
    x=[0.24062909 0.22053148] 标签:1
    x=[0.23395852 0.35842125] 标签:1
    x=[0.15146464 0.10180892] 标签:1
    x=[0.84654881 0.90576809] 标签:0
    x=[0.21555452 0.26981763] 标签:1
    x=[0.96440204 0.77809405] 标签:0
    -----
    *****
    x=[0.21995262 0.01033732] 预测为:[1]
    x=[0.78082614 0.75391697] 预测为:[0]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly96aGFuZ3BoaWwuYmxvZy5jc2RuLm5ldA_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/20191113171021338.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly96aGFuZ3BoaWwuYmxvZy5jc2RuLm5ldA==,size_16,color_FFFFFF,t_70