[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$faCrufJWEINvVDps1pRBSIByurwj-OCjdQxLiE-t88sk":3},{"answer":4,"createTime":5,"id":6,"options":7,"origin":12,"question":19,"related":20,"source":30,"type":31},[],"2024-10-04 13:31:46",160723201,[8,9,10,11],"seed = tf.random.normal(16, 100)","seed = tf.random.normal((100, 16))","seed = tf.random.normal(100, 16)","seed = tf.random.normal((16, 100))",{"count":13,"courseId":14,"courseImg":15,"courseName":16,"workId":17,"workName":18},15,"f7236923c44e2d7ea349b8dc8ea8b6c6","https:\u002F\u002Ftihai-oss-cloud.itihey.com\u002Fimg\u002Fd00903c8b3fd08da0021d53672530108.jpg","生成对抗网络原理与应用","work_36830824","2基本GAN","想要生成16个长度为100的随机向量,正确的代码应该为",[21,32,41,44,53,62,71,80,89,98],{"answer":22,"createTime":5,"id":23,"options":24,"question":29,"source":30,"type":31},[],160723199,[25,26,27,28],"生成器能够不断提高生成逼真样本的能力,而判别器则不断提高辨别真伪样本的能力","GAN 于 2014 年通过Ian Goodfellow、Yoshua Bengio 和其他蒙特利尔大学研究人员撰写的论文提出","GAN由两个主要组成部分构成:生成器(Generator)和判别器(Discriminator),这两个部分通过对抗博弈学习的方式相互竞争","生成对抗网络(GAN)是一种使用多个神经网络的计算过程,这几个神经网络被视为各自的"对手"","下列关于GAN的说法,哪一个是错误的","v1",0,{"answer":33,"createTime":5,"id":34,"options":35,"question":40,"source":30,"type":31},[],160723200,[36,37,38,39],"tf.layers.dropout(…)","tf.layers.batch_normalization(…)","tf.layers.max_pooling2d(…)","tf.layers.conv2d_transpose(…)","如果要添加批量标准化层,以下哪个代码是正确的",{"answer":42,"createTime":5,"id":6,"options":43,"question":19,"source":30,"type":31},[],[8,9,10,11],{"answer":45,"createTime":5,"id":46,"options":47,"question":52,"source":30,"type":31},[],160723202,[48,49,50,51],"求导过程,dy_dx = GradientTap.gradient(y, x),y对x求导,相当于损失函数对变量进行求导","调用生成器函数和判别器函数","计算损失函数","优化过程,apply_gradients根据generator产生的梯度来优化更新对应的variables","在下列批次优化训练函数中,最后两行代码的注释是? @tf.function def train_step(images): noise = tf.random.normal([BATCH_SIZE, noise_dim]) with tf.GradientTape() as gen_tape, tf.GradientTape() as disc_tape: generated_images = generator(noise, training=True) real_output = discriminator(images, training=True) fake_output = discriminator(generated_images, training=True) gen_loss = generator_loss(fake_output) disc_loss = discriminator_loss(real_output, fake_output) gradients_of_generator = gen_tape.gradient(gen_loss, generator.trainable_variables) gradients_of_discriminator = disc_tape.gradient(disc_loss, discriminator.trainable_variables) generator_optimizer.apply_gradients(zip(gradients_of_generator, generator.trainable_variables)) #注释 discriminator_optimizer.apply_gradients(zip(gradients_of_discriminator, discriminator.trainable_variables)) #注释",{"answer":54,"createTime":5,"id":55,"options":56,"question":61,"source":30,"type":31},[],160723203,[57,58,59,60],"@tf.function 此句代码的意思是将模型以图graph模式运行,将TensorFlow图转换为Python函数,以提高函数的执行效率","训练过程中,生成网络G的目标就是尽量生成真实的图片去欺骗判别网络D;而D的目标就是尽量把G生成的图片和真实的图片分别开来","在训练过程中,生成器和判别器通过对抗学习相互博弈,在这种对抗学习的过程中,最终两个网络达到了一个静态平衡,对于给定图像的预测为真的概率基本接近 0.5","GAN 主要用于生成新的、逼真的样本,通常在图像领域,不作用于其他如文本、音频等领域","关于GAN的工作原理,以下哪个说法是正确的",{"answer":63,"createTime":5,"id":64,"options":65,"question":70,"source":30,"type":31},[],160723204,[66,67,68,69],"zeros_like,ones_like","ones_like,ones_like","zeros_like,zeros_like","ones_like,zeros_like","下列是辨别器的损失函数,请选择正确的判定函数. cross_entropy = tf.keras.losses.BinaryCrossentropy(from_logits=True) def discriminator_loss(real_output, fake_output): real_loss = cross_entropy(tf. (real_output), real_output) fake_loss = cross_entropy(tf. (fake_output), fake_output) total_loss = real_loss + fake_loss return total_loss",{"answer":72,"createTime":5,"id":73,"options":74,"question":79,"source":30,"type":31},[],160723205,[75,76,77,78],"sigmoid函数","tanh函数","leak relu 函数","relu 函数","以下哪个激活函数,可以使得输出在[-1,1]之间",{"answer":81,"createTime":5,"id":82,"options":83,"question":88,"source":30,"type":31},[],160723206,[84,85,86,87],"第三行、第五行、第八行有错","第五行、第六行、第八行有错","第三行、第六行、第八行有错","第五行、第八行有错","判断下列生成器模型是否有错? 第一行:def generator_model(): 第二行: model = tf.keras.Sequential() 第三行: model.add(layers.Dense(512, use_bias=False, input_shape=(100))) 第四行: model.add(layers.BatchNormalization()) 第五行: model.add(layers.LeakyReLU) 第六行: model.add(layers.Dense(28*28*1, use_bias=False, activation='tanh')) 第七行: model.add(layers.BatchNormalization()) 第八行: model.add(layers.Reshape(28, 28, 1)) 第九行: return model",{"answer":90,"createTime":5,"id":91,"options":92,"question":97,"source":30,"type":31},[],160723207,[93,94,95,96],"转换类型并且归一化0到1之间","归一化-1到1之间","转换类型并且归一化-1到1之间","归一化0到1之间","train_images = (train_images - 127.5) \u002F 127.5 ,这行代码实现的功能是",{"answer":99,"createTime":5,"id":100,"options":101,"question":106,"source":30,"type":31},[],160723208,[102,103,104,105],"def generator_loss(fake_output): return cross_entropy(tf.ones_like(fake_output), fake_output)","def generator_loss(real_output, fake_output): real_loss = cross_entropy(tf.zeros_like(real_output), real_output) fake_loss = cross_entropy(tf.ones_like(fake_output), fake_output) total_loss = real_loss + fake_loss return total_loss","def generator_loss(fake_output): return cross_entropy(tf.zeros_like(fake_output), fake_output)","def generator_loss(real_output, fake_output): real_loss = cross_entropy(tf.ones_like(real_output), real_output) fake_loss = cross_entropy(tf.zeros_like(fake_output), fake_output) total_loss = real_loss + fake_loss return total_loss","编写生成器的损失函数,以下哪段代码是正确的"]