Author: Zen and the Art of Computer Programming
1 Introduction
In fields such as natural language processing, image recognition, autonomous driving, and video analysis, deep learning frameworks are an important part of modern machine learning. In recent years, a large number of researchers have applied it to various fields such as computer vision, natural language processing, autonomous driving, and medical health. In order to achieve these goals, the support of deep learning frameworks for cluster environments has become an urgent problem to be solved. Tensorflow has relatively few applications in China, and many domestic companies do not have so many resources to deploy deep learning. Therefore, this article will focus on the deployment of Tensorflow enterprise-level distributed cluster environment, combined with actual cases, to lead readers to understand how to build a Tensorflow enterprise-level cluster environment and how to use Tensorflow to achieve business needs.
2. Basic concepts and terminology
This chapter mainly introduces the basic knowledge and terminology related to deep learning to facilitate readers to understand the content to be explained in this article.
TensorFlow
TensorFlow overview
TensorFlow is an open source software library that can be used for machine learning and deep neural network research and development. It is an open source system that uses data flow graph as a computing model. The nodes in the graph represent operations and the edges represent data flow, making machine learning simpler, faster and more portable.
TensorFlow was proposed by the Google Brain team at the end of 2015 and is designed based on the concept of structured data flow graphs running on heterogeneous devices. It was originally named "TensorFlow" and later renamed "Deep Learning Software Library". Up to now, TensorFlow has released multiple versions, including version 1.x, version 2.x, and version 2.3. The latest version is currently 2.3.1.