Neural Network

In deep learning, we use artificial neural networks (ANN) as architecture to solve machine learning problems. An ANN imitates a brain by connecting perceptrons (i.e. artificial neurons) together.

Architectures

• Autoencoder: mainly used to discover and learn features from unlabeled data.
• CNN: used in computer vision and image generation
• RNN: used for sequential data (i.e. text, speech, time series)
• Transformer: architecture made of 2 main parts, a encoder and a decoder
• GAN: 2 neural networks (generator and discriminator) compete against each other

Resources

• playground.tensorflow.org – Neural network visualisation