Dcgan

Deep Convolutional Generative Adversarial Networks (DCGANs) are a class of GANs that use convolutional neural networks in both the generator and discriminator. DCGANs leverage the power of convolutional layers to generate higher quality and more realistic images compared to basic GANs.

Key characteristics of DCGANs include:

• Use of convolutional and transposed convolutional layers instead of fully connected layers.
• Removal of pooling layers; instead, strided convolutions are used for downsampling and upsampling.
• Use of batch normalization to stabilize training.
• Use of ReLU activation in the generator and LeakyReLU in the discriminator.

Basic architecture example of a DCGAN generator and discriminator using TensorFlow/Keras:

import tensorflow as tf
from tensorflow.keras import layers

# DCGAN Generator
def build_dcgan_generator():
    model = tf.keras.Sequential([
        layers.Dense(7*7*256, use_bias=False, input_shape=(100,)),
        layers.BatchNormalization(),
        layers.ReLU(),
        layers.Reshape((7, 7, 256)),
        layers.Conv2DTranspose(128, (5,5), strides=(1,1), padding='same', use_bias=False),
        layers.BatchNormalization(),
        layers.ReLU(),
        layers.Conv2DTranspose(64, (5,5), strides=(2,2), padding='same', use_bias=False),
        layers.BatchNormalization(),
        layers.ReLU(),
        layers.Conv2DTranspose(1, (5,5), strides=(2,2), padding='same', use_bias=False, activation='tanh')
    ])
    return model

# DCGAN Discriminator
def build_dcgan_discriminator():
    model = tf.keras.Sequential([
        layers.Conv2D(64, (5,5), strides=(2,2), padding='same', input_shape=[28,28,1]),
        layers.LeakyReLU(alpha=0.2),
        layers.Dropout(0.3),
        layers.Conv2D(128, (5,5), strides=(2,2), padding='same'),
        layers.LeakyReLU(alpha=0.2),
        layers.Dropout(0.3),
        layers.Flatten(),
        layers.Dense(1, activation='sigmoid')
    ])
    return model