Author: Zen and the Art of Computer Programming
1 Introduction
Deep Learning is a booming discipline in recent years. It uses computers to process big data, realize natural language understanding, image recognition and other fields, and has achieved great success. At present, deep learning has become an important research direction in the field of artificial intelligence. Its core algorithm is a neural network, which can improve the performance of the model by training massive amounts of data. Deep learning frameworks can be divided into two broad categories: frameworks and model libraries. This article will first introduce the basic concepts and development history of deep learning; then discuss the development status and development history of deep learning frameworks; finally introduce the functions and characteristics of some commonly used deep learning frameworks and model libraries.
2. Basic concepts and terminology
2.1 Deep learning
Deep learning is a machine learning method based on multi-level nonlinear activation functions, which is a set of algorithms that improve their performance by learning data representation and pattern recognition tasks. The deep learning system consists of two parts, shallow and deep, where the shallow model can quickly and efficiently identify input features, while the deep model learns a more abstract representation by combining the output of the shallow model.
2.1.1 Concept
- Neural network: Neural network (neural network) is one of the most famous deep learning models. It consists of multiple nodes connected to each other, and learns the distribution and characteristics of data by calculating the way of transferring information. The neural network has been improved on the basis of the brain's neuron network, and a series of new techniques have been added to overcome learning difficulties and solve overfitting problems.
- Deep learning: Deep learning refers to a type of machine learning technology, which is pattern recognition and prediction based on deep neural networks. The purpose of deep learning is to give computers the ability to automatically find underlying structures and patterns from raw data.
- Model: deep learning model (deep learning model