tensorboard-sklearn数据-loss

记录sklearn数据训练时的loss值，用tensorboard可视化

三步骤：红字处

import tensorflow as tf

from sklearn.datasets import load_digits

from sklearn.cross_validation import train_test_split

from sklearn.preprocessing import LabelBinarizer

# load data

digits = load_digits()

X = digits.data

y = digits.target

y = LabelBinarizer().fit_transform(y)   # 转换格式

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.3)

def add_layer(inputs, in_size, out_size, layer_name, active_function=None):

    """

    :param inputs:

    :param in_size: 行

    :param out_size: 列 ， [行， 列] =矩阵

    :param active_function:

    :return:

    """

    with tf.name_scope('layer'):

        with tf.name_scope('weights'):

            W = tf.Variable(tf.random_normal([in_size, out_size]), name='W')  #

        with tf.name_scope('bias'):

            b = tf.Variable(tf.zeros([1, out_size]) + 0.1)  # b是一行数据，对应out_size列个数据

        with tf.name_scope('Wx_plus_b'):

            Wx_plus_b = tf.matmul(inputs, W) + b

        Wx_plus_b = tf.nn.dropout(Wx_plus_b, keep_prob=keep_prob)

        if active_function is None:

            outputs = Wx_plus_b

        else:

            outputs = active_function(Wx_plus_b)

        tf.summary.histogram(layer_name + '/outputs', outputs)  # 1.2.记录outputs值，数据直方图

        return outputs

# define placeholder for inputs to network

keep_prob = tf.placeholder(tf.float32)  # 不被dropout的数量

xs = tf.placeholder(tf.float32, [None, 64])  # 8*8

ys = tf.placeholder(tf.float32, [None, 10])

# add output layer

l1 = add_layer(xs, 64, 50, 'l1', active_function=tf.nn.tanh)

prediction = add_layer(l1, 50, 10, 'l2', active_function=tf.nn.softmax)

# the loss between prediction and really

cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys * tf.log(prediction), reduction_indices=[1]))

tf.summary.scalar('loss', cross_entropy)  # 字符串类型的标量张量，包含一个Summaryprotobuf  1.1记录标量

# training

train_step = tf.train.GradientDescentOptimizer(0.5).minimize(cross_entropy)

sess = tf.Session()

merged = tf.summary.merge_all()  # 2.把所有summary节点整合在一起,只需run一次,这儿只有cross_entropy

sess.run(tf.initialize_all_variables())

train_writer = tf.summary.FileWriter('log/train', sess.graph)  # 3.写入

test_writer = tf.summary.FileWriter('log/test', sess.graph)

# start training

for i in range(500):

    sess.run(train_step, feed_dict={xs: X_train, ys: y_train, keep_prob: 0.5})  # keep_prob训练时保留50%，防止过拟合

    if i % 50 == 0:

        # record loss

        train_result = sess.run(merged, feed_dict={xs: X_train, ys: y_train, keep_prob: 1})  # 3.1 激活 tensorboard记录保留100%的数据

        test_result = sess.run(merged, feed_dict={xs: X_test, ys: y_test, keep_prob: 1})

        train_writer.add_summary(train_result, i)

        test_writer.add_summary(test_result, i)

print("Record Finished !!!")