Neural Networks

This paper introduces the outline of a neural network in the artificial intelligence field. This discussion will provide a brief overview of the history of neural networks and the latest advances in this field. In addition, we describe some practical applications of neural networks. Introduction The main goal in the field of artificial intelligence is to build a machine with intelligence equivalent to human intelligence. The pursuit of this artificial intelligence has a long history.

Let's begin by discussing the neural network. Neural network is an artificial intelligence method based on the human brain. We know that the human brain consists of billions of interconnected neurons. The neural network trains a large amount of data and each layer of the neuron develops a task to perform input from input to input to input. These types of neural networks are the foundation of many of the greatest advances in AI in the past few years, from new outcomes in accurate medical diagnosis to victories in Google and AlphaGo.

In this course, we will understand the intuition behind the artificial neural network, apply artificial neural networks to practice, understand the intuition behind the convolution neural network, actually apply the convolution neural network, I understand the background. Intuition, actually the application of recurrent neural network understands the intuition behind the self-organizing map, actually applies self-organizing mapping, understands the intuition behind the Boltzmann machine, and actually the Boltzmann machine Apply and understand the intuition behind the auto encoder

Typical and well-studied neural networks (such as image classifiers) are considered to be the left hemisphere of neural network technology. With that in mind, it is easy to understand what the Generative Adversarial Network is. It is a correct hemisphere - it is said to be responsible for the hemisphere of creativity. Generating a confrontational network (GAN) is the first step in learning neural network technology. A typical GAN ​​is a trained neural network that uses image data sets and some random noise as seeds to generate images of specific subjects. So far, GAN created images with low quality and limited resolution. Recent developments at NVIDIA show that it is achievable to generate realistic images with high resolution and they released the technology itself on open access.