Algorithms and Architectures of Artificial Intelligence

Today, if you want to be a computer scientist, you went to college and was trained in algorithms, data structures, computer architecture, artificial intelligence etc. You will learn the programming language. You will learn how to debug your code. You will learn how to build the system. To wait

Artificial intelligence requires appropriate machine learning algorithms, data and computational power. Most of today's most promising artificial intelligence techniques use neural networks. Neural networks can be small and simple, or large and powerful. The larger the neural network, the stronger it is. To achieve the limits of neural networks, we need to use powerful computer chips as well as today's artificial intelligence functions. Intel has long been a market leader in supercomputer chips, but Nvidia, Google and other companies are trying to develop chips, especially for deep learning and other artificial intelligence.

The scale of artificial neural networks, especially the deep learning model, is expanding every year. The higher these algorithms and architectures, the more data you can process. However, the hardware design is difficult to catch up, which raises the question of how to improve chip calculation capability while keeping the chip compact. According to Moore's Law, the throughput of semiconductor computing is only doubled every 18 months. Other hardware improvements must rely on new architecture and optimization