Code fragments of rainGAN.py

• import os
• from os import path
• import numpy as np
• from tensorflow.keras.optimizers import Adam
• from tensorflow.keras.layers import Input, Dense, Reshape, Flatten, Conv2D, UpSampling2D, BatchNormalization, MaxPooling2D, Activation
• from tensorflow.keras.models import Sequential, Model
• lr = 0.0002
• batch_size = 128
• optimizer = Adam(lr=lr)
• def generator():
• model = Sequential()
• model.add(Dense(1024, input_shape=(100,)))
• model.add(BatchNormalization(momentum=0.8))
• model.add(Activation('tanh'))
• model.add(Reshape((2, 2, 256), input_shape=(1024,)))
• model.add(UpSampling2D(size=(2, 2)))
• model.add(Conv2D(512, (5, 5), padding="same"))
• model.add(BatchNormalization(momentum=0.8))
• model.add(Activation('tanh'))
• model.add(UpSampling2D(size=(2, 2)))
• model.add(Conv2D(256, (5, 5), padding="same"))
• model.add(Activation('tanh'))
• model.add(UpSampling2D(size=(2, 2)))
• model.add(Conv2D(128, (5, 5), padding="same"))
• model.add(Activation('tanh'))
• model.add(UpSampling2D(size=(2, 2)))
• model.add(Conv2D(1, (5, 5), padding='same'))
• model.add(Activation('tanh'))
• noise = Input((100,))
• generated = model(noise)
• return Model(noise, generated)
• def discriminator():
• model = Sequential()
• model.add(Conv2D(64, (5, 5), padding='same', input_shape=(32, 32, 1)))
• model.add(Activation('tanh'))
• model.add(MaxPooling2D(pool_size=(2, 2)))
• model.add(Conv2D(128, (5, 5), padding='same'))
• model.add(Activation('tanh'))
• model.add(MaxPooling2D(pool_size=(2, 2)))
• model.add(Conv2D(256, (5, 5), padding='same'))
• model.add(Activation('tanh'))
• model.add(MaxPooling2D(pool_size=(2, 2)))
• model.add(Conv2D(512, (5, 5), padding='same'))
• model.add(Activation('tanh'))
• model.add(MaxPooling2D(pool_size=(2, 2)))
• model.add(Flatten())
• model.add(Dense(1))
• model.add(Activation('sigmoid'))
• data = Input((32, 32, 1))
• validity = model(data)
• return Model(data, validity)
• def get_avg_weights(array_weights):
• new_weights = list()
• for weights_list_tuple in zip(*array_weights):
• new_weights.append( np.array([np.array(w).mean(axis=0) for w in zip(*weights_list_tuple)]) )
• return new_weights
• def build_models():
• dis = discriminator()
• dis.compile(loss='binary_crossentropy', optimizer=optimizer, metrics=['accuracy'])
• gen = generator()
• dis.trainable = False
• z = Input((100,))
• img = gen(z)
• valid = dis(img)
• combined = Model(z, valid)
• combined.compile(loss='binary_crossentropy', optimizer=optimizer)
• dis.trainable = True
• return combined, dis, gen
• def compile_models(dis, gen):
• dis.compile(loss='binary_crossentropy', optimizer=optimizer, metrics=['accuracy'])
• dis.trainable = False
• z = Input((100,))
• img = gen(z)
• valid = dis(img)
• combined = Model(z, valid)
• combined.compile(loss='binary_crossentropy', optimizer=optimizer)
• dis.trainable = True
• return combined, dis, gen
• def ciil(combined, dis, gen, epochs_per_cycle, dataset, min_acc):
• normalise_dataset(dataset)
• c_epoch = 0
• for epoch in range(epochs_per_cycle):
• c_epoch = epoch_procedure(c_epoch, combined, dataset, dis, gen, min_acc)
• print("Cycle: [" + str(epoch + 1) + "/" + str(epochs_per_cycle) + "] complete.")
• save(c_epoch, dis, gen, combined)
• def normalise_dataset(dataset):
• for i in range(len(dataset)):
• dataset[i] = (dataset[i].astype(np.float32) - 2047.5) / 2047.5
• def epoch_procedure(c_epoch, combined, dataset, dis, gen, min_acc):
• c_epoch += 1
• d_loss = [0, 0]
• while d_loss[1] < min_acc:
• rand = np.random.randint(0, dataset.shape[0], batch_size // 2)
• trainset = dataset[rand]
• noise = np.random.normal(0, 1, (batch_size // 2, 100))
• generated = gen.predict(noise)
• d_loss_real = dis.train_on_batch(trainset, np.ones((batch_size // 2, 1)))
• d_loss_fake = dis.train_on_batch(generated, np.zeros((batch_size // 2, 1)))
• d_loss = 0.5 * np.add(d_loss_real, d_loss_fake)
• noise = np.random.normal(0, 1, (batch_size, 100))
• valid_y = np.array([1] * batch_size)
• g_loss = combined.train_on_batch(noise, valid_y)
• return c_epoch
• def save(dis, gen, combined):
• print("Saving Generator Instance...")
• combined.save("./models/modelCombined.h5")
• dis.save("./models/modelDis.h5")
• gen.save("./models/modelGen.h5")

Graph