Community Blog What is Deep Learning

What is Deep Learning

Deep learning is a new research direction in the field of machine learning. It was introduced into machine learning to make it closer to the original goal, artificial intelligence.

Deep Learning Definition

Deep learning is to learn the inherent laws and representation levels of sample data. The information obtained in the learning process is of great help to the interpretation of data such as text, images, and sounds. Its ultimate goal is to enable machines to have the ability to analyze and learn like humans, and to recognize data such as text, images, and sounds. Deep learning is a complex machine learning algorithm that has achieved results in speech and image recognition far beyond previous related technologies.

Deep learning has achieved many results in search technology, data mining, machine learning, machine translation, natural language processing, multimedia learning, speech, recommendation and personalization technology, and other related fields. Deep learning enables machines to imitate human activities such as audio-visual and thinking and solves many complex pattern recognition problems, which has made great progress in artificial intelligence-related technologies.

Deep learning is a type of machine learning, and machine learning is the necessary path to realize artificial intelligence. The concept of deep learning originates from the research of artificial neural networks, and a multilayer perceptron with multiple hidden layers is a deep learning structure. Deep learning forms a more abstract high-level representation attribute category or feature by combining low-level features to discover distributed feature representations of data. The motivation for studying deep learning is to build a neural network that simulates the human brain for analysis and learning. It mimics the mechanism of the human brain to interpret data, such as images, sounds, and text.

Deep Learning Feature

Different from traditional shallow learning, deep learning is:

  • The depth of the model structure is emphasized, usually with 5 layers, 6 layers, or even 10 layers of hidden nodes;
  • The importance of feature learning is clarified. In other words, by layer-by-layer feature transformation, the feature representation of the sample in the original space is transformed into a new feature space, thereby making classification or prediction easier. Compared with the method of constructing features by artificial rules, the use of big data to learn features is more capable of portraying data-rich internal information;

Through the design and establishment of an appropriate number of neuron computing nodes and multi-layer computing hierarchy, select the appropriate input layer and output layer. Through the learning and tuning of the network, a functional relationship from input to output is established. Although the functional relationship between input and output cannot be found 100%, the actual relationship can be approximated as much as possible. Using a successfully trained network model, we can achieve our automation requirements for complex transaction processing.

Typical Deep Learning Model

Typical deep learning models include convolutional neural network, DBN, and stacked auto-encoder network models, etc.

  • Convolutional Neural Network Model

Before the advent of unsupervised pre-training, training deep neural networks were usually very difficult, and one of the special cases was convolutional neural networks. Convolutional neural networks are inspired by the structure of the visual system.
The first convolutional neural network calculation model was proposed in Fukushima's neurocognitive machine.
Based on the local connection between neurons and the hierarchical organization image conversion, the neurons with the same parameters are applied to different positions of the previous layer of the neural network, and a translation-invariant neural network structure is obtained. Later, based on this idea, Le Cun used error gradients to design and train convolutional neural networks to obtain superior performance on some pattern recognition tasks. So far, the pattern recognition system based on a convolutional neural network is one of the best implementation systems, especially in handwritten character recognition tasks that show extraordinary performance.

  • Deep Belief Network (DBN) Model

DBN can be interpreted as a Bayesian probability generation model. It is composed of multiple layers of random hidden variables. The upper two layers have undirected symmetrical connections, and the lower layer gets top-down directed connections from the upper layer. The state of the lowest layer unit is the visible input data vector. DBN is composed of a stack of 2F structural units. The structural unit is usually RBM (RestIlcted Boltzmann Machine). The number of neurons in the visible layer of each RBM unit in the stack is equal to the number of neurons in the hidden layer of the previous RBM unit. According to the deep learning mechanism, the input samples are used to train the first-layer RBM units, and their output is used to train the second-layer RBM models, and the RBM models are stacked to improve the model performance by adding layers. In the unsupervised pre-training process, after the DBN code is input to the top RBM, the state of the top layer is decoded to the unit of the bottom layer to realize the reconstruction of the input. As the structural unit of DBN, RBM shares parameters with each layer of DBN.

  • Stacked Auto-Encoder Network Model

The structure of the stacked auto-encoding network is similar to that of the DBN, consisting of a stack of several structural units. The difference is that the structural unit is an auto-en-coder instead of RBM. The self-encoding model is a two-layer neural network, the first layer is called the coding layer, and the second layer is called the decoding layer.

Deep Learning Application

  • Computer vision
    The Multimedia Laboratory of the Chinese University of Hong Kong is the first Chinese team to apply deep learning for computer vision research. In the world-class artificial intelligence competition LFW (Large-scale Face Recognition Competition), the laboratory once beat FaceBook to win the championship, making the recognition ability of artificial intelligence in this field surpasses that of real people for the first time.
  • Speech Recognition
    Through cooperation with hinton, Microsoft researchers first introduced RBM and DBN into the training of speech recognition acoustic models and achieved great success in large vocabulary speech recognition systems, which resulted in a relatively 30% reduction in the error rate of speech recognition. However, DNN does not yet have effective parallel fast algorithms. Many research institutions are using large-scale data corpus to improve the training efficiency of DNN acoustic models through the GPU platform.
  • Natural language processing and other fields
    Deep learning is mainly used in machine translation and semantic mining in fields such as natural language processing.

Related Blog

AI vs Machine Learning vs Deep Learning

The wave of artificial intelligence is sweeping the world, and many words linger in our ears all the time: artificial intelligence(AI), machine learning, and deep learning. Many people always seem to understand the meaning of these high-frequency words and the relationship behind them.

To better understand artificial intelligence(AI), this article explains the meaning of these words in the simplest language and clarifies the relationship between them, hoping to be helpful to people who are just getting started.

What Is Machine Learning

Machine Learning (ML) in simple terms can be defined as the science of getting computers to act and learn without explicit programming to perform those actions. It has become quite popular in recent years, however, the term itself was coined in 1959 by Arthur Samuel who defined Machine Learning as ‘the field of study that gives computers the ability to learn without being explicitly taught’.

A more recent and formal definition of Machine Learning was created by Tom Mitchell and describes it as a well-defined learning problem – ‘A computer program is said to learn from experience E concerning some class of tasks T and performance measure P if its performance at tasks in T, as measured by P, improves with experience E.

6 Mistakes To Avoid While Building Your Machine Learning Model

In recent years, machine learning has received more and more attention in the field of academic research and practical applications. But building a machine learning model is not a simple matter. It requires a lot of knowledge and skills and rich experience to make the model work in a variety of scenarios. The correct machine learning model should be data-centric and based on an understanding of business problems. And data and machine learning algorithms must be applied to solve problems to build a machine learning model that can meet the needs of the project.

Related Product

Machine Learning Platform for AI

Machine Learning Platform for AI provides end-to-end machine learning services, including data processing, feature engineering, model training, model prediction, and model evaluation. Machine Learning Platform for AI combines all of these services to make AI more accessible than ever.

Alibaba Cloud Campaign

Retail Innovation Summit Europe

This half-day online conference will help retail and e-commerce leaders to better address new challenges presented in the digital era under different business scenarios. You can also have a live chat with our experts to benefit from our leading technologies that fuel Alibaba's e-commerce business.

0 0 0
Share on

Alibaba Clouder

2,600 posts | 750 followers

You may also like

Alibaba Clouder

2,600 posts | 750 followers

Related Products