ABSTRACT
Different types of neural network architectures exist, though they are generally based on neural networks with multiple layers of stacked neurons that allows the backward propagation of a signal. In general, neural networks can be viewed as generative or discriminative models where discriminative model discriminates between different kinds of data instances, while generative models can generate new data instances. In terms of network learning mode, two of the commonest deep learning architectures include systems based on supervised end-to-end training and systems based on unsupervised, layer-by-layer pre-training of networks, with supervised fine-tuning of the network.
Most pre-trained neural network models in deep learning can be broadly grouped into artificial neural networks, convolutional neural networks and recurrent neural networks. However, there are also neural networks that do not fall directly into any of these classes because they operate using an integration of multiple network types such as generative adversarial networks and deep belief network. As neural network types continue to grow exponentially, keeping up with the various types of architectures that are continually emerging gets more difficult. This chapter discusses the modern neural network architectures that are frequently used in various applications. [190 words]
