infoGAN.py

#-*- coding: utf-8 -*-
from __future__ import division
import os
import time
import tensorflow as tf
import numpy as np

from ops import *
from utils import *

class infoGAN(object):
    # SUPERVISED是设置起来用来决定是否利用标签来训练模型
    def __init__(self, sess, epoch, batch_size, z_dim, dataset_name, checkpoint_dir, result_dir, log_dir, SUPERVISED=False):
        self.sess = sess
        self.dataset_name = dataset_name
        self.checkpoint_dir = checkpoint_dir
        self.result_dir = result_dir
        self.log_dir = log_dir
        self.epoch = epoch
        self.batch_size = batch_size
        self.model_name = "infoGAN"     # name for checkpoint

        if dataset_name == 'mnist' or dataset_name == 'fashion-mnist':
            # parameters
            self.input_height = 28
            self.input_width = 28
            self.output_height = 28
            self.output_width = 28

            self.z_dim = z_dim         # dimension of noise-vector
            self.y_dim = 12         # dimension of code-vector (label+two features)为什么采取12个呢，除了类别外加了两个特征
            self.c_dim = 1

            self.SUPERVISED = SUPERVISED # if it is true, label info is directly used for code

            # train
            self.learning_rate = 0.0002
            self.beta1 = 0.5

            # test
            self.sample_num = 64  # number of generated images to be saved

            # code
            self.len_discrete_code = 10  # categorical distribution (i.e. label) 类别
            self.len_continuous_code = 2  # gaussian distribution (e.g. rotation, thickness) 旋转，字体

            # load mnist
            self.data_X, self.data_y = load_mnist(self.dataset_name)

            # get number of batches for a single epoch
            self.num_batches = len(self.data_X) // self.batch_size
        else:
            raise NotImplementedError

    # 分类器,实现(64,x)-->(64,64)-->(64,12)的变换，确定分类结果
    def classifier(self, x, is_training=True, reuse=False):
        # Network Architecture is exactly same as in infoGAN (https://arxiv.org/abs/1606.03657)
        # Architecture : (64)5c2s-(128)5c2s_BL-FC1024_BL-FC128_BL-FC12S’
        # All layers except the last two layers are shared by discriminator
        # Number of nodes in the last layer is reduced by half. It gives better results.
        with tf.variable_scope("classifier", reuse=reuse):

            net = lrelu(bn(linear(x, 64, scope='c_fc1'), is_training=is_training, scope='c_bn1'))
            out_logit = linear(net, self.y_dim, scope='c_fc2')
            out = tf.nn.softmax(out_logit)

            return out, out_logit

    # 判别器与之前模型的作用完全一样
    def discriminator(self, x, is_training=True, reuse=False):
        # Network Architecture is exactly same as in infoGAN (https://arxiv.org/abs/1606.03657)
        # Architecture : (64)4c2s-(128)4c2s_BL-FC1024_BL-FC1_S
        with tf.variable_scope("discriminator", reuse=reuse):

            net = lrelu(conv2d(x, 64, 4, 4, 2, 2, name='d_conv1'))
            net = lrelu(bn(conv2d(net, 128, 4, 4, 2, 2, name='d_conv2'), is_training=is_training, scope='d_bn2'))
            net = tf.reshape(net, [self.batch_size, -1])
            net = lrelu(bn(linear(net, 1024, scope='d_fc3'), is_training=is_training, scope='d_bn3'))
            out_logit = linear(net, 1, scope='d_fc4')
            out = tf.nn.sigmoid(out_logit)

            return out, out_logit, net

    # 生成器与CGAN一样，用于处理带标签的生成
    def generator(self, z, y, is_training=True, reuse=False):
        # Network Architecture is exactly same as in infoGAN (https://arxiv.org/abs/1606.03657)
        # Architecture : FC1024_BR-FC7x7x128_BR-(64)4dc2s_BR-(1)4dc2s_S
        with tf.variable_scope("generator", reuse=reuse):

            # merge noise and code
            z = concat([z, y], 1)

            net = tf.nn.relu(bn(linear(z, 1024, scope='g_fc1'), is_training=is_training, scope='g_bn1'))
            net = tf.nn.relu(bn(linear(net, 128 * 7 * 7, scope='g_fc2'), is_training=is_training, scope='g_bn2'))
            net = tf.reshape(net, [self.batch_size, 7, 7, 128])
            net = tf.nn.relu(
                bn(deconv2d(net, [self.batch_size, 14, 14, 64], 4, 4, 2, 2, name='g_dc3'), is_training=is_training,
                   scope='g_bn3'))

            out = tf.nn.sigmoid(deconv2d(net, [self.batch_size, 28, 28, 1], 4, 4, 2, 2, name='g_dc4'))

            return out

    def build_model(self):
        # some parameters
        image_dims = [self.input_height, self.input_width, self.c_dim]
        bs = self.batch_size

        """ Graph Input """
        # images
        self.inputs = tf.placeholder(tf.float32, [bs] + image_dims, name='real_images')

        # labels
        self.y = tf.placeholder(tf.float32, [bs, self.y_dim], name='y')

        # noises
        self.z = tf.placeholder(tf.float32, [bs, self.z_dim], name='z')

        """ Loss Function """
        ## 1. GAN Loss
        # output of D for real images
        D_real, D_real_logits, _ = self.discriminator(self.inputs, is_training=True, reuse=False)

        # 取鉴别器的最后一层作为类别信息的判断
        # output of D for fake images
        G = self.generator(self.z, self.y, is_training=True, reuse=False)
        D_fake, D_fake_logits, input4classifier_fake = self.discriminator(G, is_training=True, reuse=True)

        # infoGAN的损失函数可以分为3块，D，G和C，难点在于损失C的求解
        # get loss for discriminator
        d_loss_real = tf.reduce_mean(
            tf.nn.sigmoid_cross_entropy_with_logits(logits=D_real_logits, labels=tf.ones_like(D_real)))
        d_loss_fake = tf.reduce_mean(
            tf.nn.sigmoid_cross_entropy_with_logits(logits=D_fake_logits, labels=tf.zeros_like(D_fake)))

        self.d_loss = d_loss_real + d_loss_fake

        # get loss for generator
        self.g_loss = tf.reduce_mean(
            tf.nn.sigmoid_cross_entropy_with_logits(logits=D_fake_logits, labels=tf.ones_like(D_fake)))

        # 此步用于确定分类后的类别信息
        ## 2. Information Loss
        code_fake, code_logit_fake = self.classifier(input4classifier_fake, is_training=True, reuse=False)

        # discrete code : categorical
        # 记录前十个信息量为mnist数字类别
        disc_code_est = code_logit_fake[:, :self.len_discrete_code]
        # 记录标签的类别,分为有监督和无监督两类，由SUPERVISED决定
        disc_code_tg = self.y[:, :self.len_discrete_code]
        # q_disc_loss=disc_code_tg*(-log(sigmoid(disc_code_est)))+(1-disc_code_tg)*(-log(1-sigmoid(disc_code_est)))
        q_disc_loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=disc_code_est, labels=disc_code_tg))

        # continuous code : gaussian
        # 计算除交叉商的高斯函数部分损失，用于完善q_loss
        cont_code_est = code_fake[:, self.len_discrete_code:]
        cont_code_tg = self.y[:, self.len_discrete_code:]
        q_cont_loss = tf.reduce_mean(tf.reduce_sum(tf.square(cont_code_tg - cont_code_est), axis=1))

        # get information loss
        # 类别损失与特征损失求和
        self.q_loss = q_disc_loss + q_cont_loss

        """ Training """
        # divide trainable variables into a group for D and a group for G
        t_vars = tf.trainable_variables()
        d_vars = [var for var in t_vars if 'd_' in var.name]
        g_vars = [var for var in t_vars if 'g_' in var.name]
        q_vars = [var for var in t_vars if ('d_' in var.name) or ('c_' in var.name) or ('g_' in var.name)]

        # optimizers
        # 同时训练三个损失函数达到最小
        with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
            self.d_optim = tf.train.AdamOptimizer(self.learning_rate, beta1=self.beta1) \
                .minimize(self.d_loss, var_list=d_vars)
            self.g_optim = tf.train.AdamOptimizer(self.learning_rate * 5, beta1=self.beta1) \
                .minimize(self.g_loss, var_list=g_vars)
            self.q_optim = tf.train.AdamOptimizer(self.learning_rate * 5, beta1=self.beta1) \
                .minimize(self.q_loss, var_list=q_vars)

        """" Testing """
        # for test
        self.fake_images = self.generator(self.z, self.y, is_training=False, reuse=True)

        """ Summary """
        d_loss_real_sum = tf.summary.scalar("d_loss_real", d_loss_real)
        d_loss_fake_sum = tf.summary.scalar("d_loss_fake", d_loss_fake)
        d_loss_sum = tf.summary.scalar("d_loss", self.d_loss)
        g_loss_sum = tf.summary.scalar("g_loss", self.g_loss)

        q_loss_sum = tf.summary.scalar("g_loss", self.q_loss)
        q_disc_sum = tf.summary.scalar("q_disc_loss", q_disc_loss)
        q_cont_sum = tf.summary.scalar("q_cont_loss", q_cont_loss)

        # final summary operations
        self.g_sum = tf.summary.merge([d_loss_fake_sum, g_loss_sum])
        self.d_sum = tf.summary.merge([d_loss_real_sum, d_loss_sum])
        self.q_sum = tf.summary.merge([q_loss_sum, q_disc_sum, q_cont_sum])

    def train(self):

        # initialize all variables
        tf.global_variables_initializer().run()

        # graph inputs for visualize training results
        self.sample_z = np.random.uniform(-1, 1, size=(self.batch_size , self.z_dim))
        self.test_labels = self.data_y[0:self.batch_size]
        self.test_codes = np.concatenate((self.test_labels, np.zeros([self.batch_size, self.len_continuous_code])),
                                           axis=1)

        # saver to save model
        self.saver = tf.train.Saver()

        # summary writer
        self.writer = tf.summary.FileWriter(self.log_dir + '/' + self.model_name, self.sess.graph)

        # restore check-point if it exits
        could_load, checkpoint_counter = self.load(self.checkpoint_dir)
        if could_load:
            start_epoch = (int)(checkpoint_counter / self.num_batches)
            start_batch_id = checkpoint_counter - start_epoch * self.num_batches
            counter = checkpoint_counter
            print(" [*] Load SUCCESS")
        else:
            start_epoch = 0
            start_batch_id = 0
            counter = 1
            print(" [!] Load failed...")

        # loop for epoch
        start_time = time.time()
        for epoch in range(start_epoch, self.epoch):

            # get batch data
            for idx in range(start_batch_id, self.num_batches):
                batch_images = self.data_X[idx*self.batch_size:(idx+1)*self.batch_size]

                # generate code
                if self.SUPERVISED == True:
                    batch_labels = self.data_y[idx * self.batch_size:(idx + 1) * self.batch_size]
                else:
                    # 如果是无监督学习的话，自动生成随机标签信息
                    # np.random.multinomial多项式分布，本处为10次事件中发生事件的概率为1/10，则10次中仅有一次发生而且是完全随机的
                    batch_labels = np.random.multinomial(1,
                                                         self.len_discrete_code * [float(1.0 / self.len_discrete_code)],
                                                         size=[self.batch_size])
                #另外两个特征信息完全由均值分布决定
                batch_codes = np.concatenate((batch_labels, np.random.uniform(-1, 1, size=(self.batch_size, 2))),
                                             axis=1)

                batch_z = np.random.uniform(-1, 1, [self.batch_size, self.z_dim]).astype(np.float32)

                # update D network
                _, summary_str, d_loss = self.sess.run([self.d_optim, self.d_sum, self.d_loss],
                                                       feed_dict={self.inputs: batch_images, self.y: batch_codes,
                                                                  self.z: batch_z})
                self.writer.add_summary(summary_str, counter)

                # update G and Q network
                _, summary_str_g, g_loss, _, summary_str_q, q_loss = self.sess.run(
                    [self.g_optim, self.g_sum, self.g_loss, self.q_optim, self.q_sum, self.q_loss],
                    feed_dict={self.inputs: batch_images, self.z: batch_z, self.y: batch_codes})
                self.writer.add_summary(summary_str_g, counter)
                self.writer.add_summary(summary_str_q, counter)

                # display training status
                counter += 1
                if np.mod(counter, 50) == 0:
                    print("Epoch: [%2d] [%4d/%4d] time: %4.4f, d_loss: %.8f, g_loss: %.8f" \
                          % (epoch, idx, self.num_batches, time.time() - start_time, d_loss, g_loss))

                # save training results for every 300 steps
                if np.mod(counter, 300) == 0:
                    # 只生成类别与另外两个特征信息无关
                    samples = self.sess.run(self.fake_images,
                                            feed_dict={self.z: self.sample_z, self.y: self.test_codes})
                    tot_num_samples = min(self.sample_num, self.batch_size)
                    manifold_h = int(np.floor(np.sqrt(tot_num_samples)))
                    manifold_w = int(np.floor(np.sqrt(tot_num_samples)))
                    save_images(samples[:manifold_h * manifold_w, :, :, :], [manifold_h, manifold_w],
                                './' + check_folder(self.result_dir + '/' + self.model_dir) + '/' + self.model_name
                                + '_train_{:02d}_{:04d}.png'.format(epoch, idx))

            # After an epoch, start_batch_id is set to zero
            # non-zero value is only for the first epoch after loading pre-trained model
            start_batch_id = 0

            # save model
            self.save(self.checkpoint_dir, counter)

            # show temporal results
            self.visualize_results(epoch)

        # save model for final step
        self.save(self.checkpoint_dir, counter)

    # 本函数用于验证infoGAN的实现
    def visualize_results(self, epoch):
        tot_num_samples = min(self.sample_num, self.batch_size)
        image_frame_dim = int(np.floor(np.sqrt(tot_num_samples)))

        """ random noise, random discrete code, fixed continuous code """
        # 创建随机类别标签（64,10）
        y = np.random.choice(self.len_discrete_code, self.batch_size)
        y_one_hot = np.zeros((self.batch_size, self.y_dim))
        y_one_hot[np.arange(self.batch_size), y] = 1

        z_sample = np.random.uniform(-1, 1, size=(self.batch_size, self.z_dim))

        # 仅显示类别
        samples = self.sess.run(self.fake_images, feed_dict={self.z: z_sample, self.y: y_one_hot})

        save_images(samples[:image_frame_dim * image_frame_dim, :, :, :], [image_frame_dim, image_frame_dim],
                    check_folder(self.result_dir + '/' + self.model_dir) + '/' + self.model_name + '_epoch%03d'
                    % epoch + '_test_all_classes.png')

        """ specified condition, random noise """
        n_styles = 10  # must be less than or equal to self.batch_size

        np.random.seed()
        si = np.random.choice(self.batch_size, n_styles)

        # 此处仅显示正常的图像类别
        for l in range(self.len_discrete_code):
            y = np.zeros(self.batch_size, dtype=np.int64) + l
            y_one_hot = np.zeros((self.batch_size, self.y_dim))
            y_one_hot[np.arange(self.batch_size), y] = 1

            samples = self.sess.run(self.fake_images, feed_dict={self.z: z_sample, self.y: y_one_hot})
            save_images(samples[:image_frame_dim * image_frame_dim, :, :, :], [image_frame_dim, image_frame_dim],
                        check_folder(self.result_dir + '/' + self.model_dir) + '/' + self.model_name +
                        '_epoch%03d' % epoch + '_test_class_%d.png' % l)

            samples = samples[si, :, :, :]

            if l == 0:
                all_samples = samples
            else:
                all_samples = np.concatenate((all_samples, samples), axis=0)

        """ save merged images to check style-consistency """
        canvas = np.zeros_like(all_samples)
        for s in range(n_styles):
            for c in range(self.len_discrete_code):
                canvas[s * self.len_discrete_code + c, :, :, :] = all_samples[c * n_styles + s, :, :, :]

        save_images(canvas, [n_styles, self.len_discrete_code],
                    check_folder(self.result_dir + '/' + self.model_dir) + '/' + self.model_name + '_epoch%03d' % epoch
                    + '_test_all_classes_style_by_style.png')

        """ fixed noise """
        assert self.len_continuous_code == 2

        c1 = np.linspace(-1, 1, image_frame_dim)
        c2 = np.linspace(-1, 1, image_frame_dim)
        xv, yv = np.meshgrid(c1, c2)
        xv = xv[:image_frame_dim, :image_frame_dim]
        yv = yv[:image_frame_dim, :image_frame_dim]

        c1 = xv.flatten()
        c2 = yv.flatten()

        z_fixed = np.zeros([self.batch_size, self.z_dim])

        for l in range(self.len_discrete_code):
            y = np.zeros(self.batch_size, dtype=np.int64) + l
            y_one_hot = np.zeros((self.batch_size, self.y_dim))
            y_one_hot[np.arange(self.batch_size), y] = 1

            y_one_hot[np.arange(image_frame_dim*image_frame_dim), self.len_discrete_code] = c1
            y_one_hot[np.arange(image_frame_dim*image_frame_dim), self.len_discrete_code+1] = c2

            samples = self.sess.run(self.fake_images,
                                    feed_dict={ self.z: z_fixed, self.y: y_one_hot})

            save_images(samples[:image_frame_dim * image_frame_dim, :, :, :], [image_frame_dim, image_frame_dim],
                        check_folder(self.result_dir + '/' + self.model_dir) + '/' + self.model_name +
                        '_epoch%03d' % epoch + '_test_class_c1c2_%d.png' % l)

    @property
    # 加载创建固定模型下的路径，本处为CGAN下的训练
    def model_dir(self):
        return "{}_{}_{}_{}".format(
            self.model_name, self.dataset_name,
            self.batch_size, self.z_dim)

    # 本函数的目的是在于保存训练模型后的checkpoint
    def save(self, checkpoint_dir, step):
        checkpoint_dir = os.path.join(checkpoint_dir, self.model_dir, self.model_name)

        if not os.path.exists(checkpoint_dir):
            os.makedirs(checkpoint_dir)

        self.saver.save(self.sess,os.path.join(checkpoint_dir, self.model_name+'.model'), global_step=step)

    # 本函数的意义在于读取训练好的模型参数的checkpoint
    def load(self, checkpoint_dir):
        import re
        print(" [*] Reading checkpoints...")
        checkpoint_dir = os.path.join(checkpoint_dir, self.model_dir, self.model_name)

        ckpt = tf.train.get_checkpoint_state(checkpoint_dir)
        if ckpt and ckpt.model_checkpoint_path:
            ckpt_name = os.path.basename(ckpt.model_checkpoint_path)
            self.saver.restore(self.sess, os.path.join(checkpoint_dir, ckpt_name))
            counter = int(next(re.finditer("(\d+)(?!.*\d)",ckpt_name)).group(0))
            print(" [*] Success to read {}".format(ckpt_name))
            return True, counter
        else:
            print(" [*] Failed to find a checkpoint")
            return False, 0