Là gì

Những Thông Tin Về Cấu Trúc Mạng CNN Là Gì?

Article Information About What is CNN Network Structure? belongs to the topic of Answering the Question that is being interested by a lot of you right !! Today, let’s find out with yt2byt.edu.vn The Information About What is CNN Network Structure? in today’s post!

You are viewing content: “What is CNN Network Structure Information?”

What is CNN – This is a question that many people are looking for today. This is one of the words that is very familiar to those who follow programming or technology-related professions. And if you are intending to challenge yourself in this field, do not ignore the knowledge that ITNavi shares in the following article.

What is the definition of CNN?

CNN is an abbreviation of the word Convolutional Neural Network (also known as CNNs_carrying convolutional neurons). This is one of the most advanced Deep Learning models. CNN will allow you to build intelligent systems with extremely high accuracy. Currently, CNN is widely applied in object recognition problems in images. And detailed knowledge of CNN has been explained as follows:



This is a type of sliding window located on a matrix. Convolutional layers will have learned parameters to adjust and get the most accurate information without having to select features. Convolution or convolution is the multiplication of elements in a matrix. Sliding Window is also known as kernel, filter or feature detect and is a small matrix type.

What is CNN Network?



Feature is a feature, CNNs will compare the image piece by piece and these fragments are called Feature. Instead of having to match the images together, CNN will see the similarity when raw searching for features that match with two better images. Each Feature is considered a mini image which means they are small 2D arrays. These features all correspond to certain aspects of the image and they are likely to fit together.

Readers refer to:


Recruiting python HCM with high salary and attractive regime

Recruitment of high-paying ios jobs with attractive regimes

Hiring attractive high salary php

What are the basic layers of a CNN network?

CNN includes the following basic layers:

Convolutional layers

This is the most important layer of CNN, this layer is responsible for performing all calculations. The important elements of a convolutional layer are: stride, padding, filter map, feature map.

  • CNN uses filters to apply to areas of the image. These filter maps are called 3-dimensional matrices, inside of which are numbers and they are parameters.
  • Stride means when you shift the filter map by pixels based on the total value minus left to right. And this shift is Stride.
  • Padding: The sums of 0 values ​​added to the input class.
  • Feature map: It shows the result of each time the filter map scans the input. After each scan will happen the calculation process.

Relu Layers

Relu layer is the activation function in neural network and this function is also known as activation function. The activation function is used to simulate neurons with pulse transmission rates through the axon. In the activation function, it also has a function that means: Relu, Leaky, Tanh, Sigmoid, Maxout, etc. Currently, the relu function is used daily and is extremely popular.

It is used a lot for training purposes of neural networks, then relu brings a lot of outstanding advantages such as: the calculation will become faster, … the process of using relu, we need to pay attention to the problem of customizing the parameters. learning rate and dead unit tracking. The relu layers were used after the filter map was calculated and the relu function was applied to the total values ​​of the filter map.

Pooling layer

When the input is too large, the pooling layers will be placed between the Convolutional layers to reduce the parameter. Currently, the pooling layer has two main types: max pooling and average.

Structure of CNN

Fully connected layer

This layer is responsible for giving the results after the convolutional layer and pooling layer have received the transmitted image. At this point, we get the result that the model has read the information of the images and to connect them and produce more output, we use a fully connected layer.

on the other hand, if fully connected layers have image data, they will convert it to unsegmented quality. This is quite similar to a vote, which is then evaluated to select the highest quality image.

Readers refer to:

What is Internet? What does the Intern do and the notes when working at this place?

What is Webpack? Quickly learn the knowledge about Webpack

What is the structure of a CNN network?

Bringing CNN is one of several sets of overlapping Convolution layers as well as using nonlinear activation functions like ReLU and tanh to activate the weights in the node. This class, after passing the function, will be weighted in the nodes. After these classes have passed the activation function, the ability to create more abstract information for the next classes.

In the CNN model there is invariant and associative integration. If you have the same object but project it from different angles, the accuracy is likely to be affected. For shifting, rotation, and scaling, the pooling layer will be used to help invariant these properties. Therefore, CNN will give results with corresponding accuracy in each model.

In it, the pooling layer will give you invariant for displacement, scaling and rotation. The local associativity will show you the low to high levels of representation, information with abstraction through convolution from the filters. The CNN model has layers that are interconnected based on the convolution mechanism.

Subsequent layers will result from convolutions from the previous layer, so you will have the best local connections. Thus, each neuron in the next layer generated from the resulting filter will be applied to the local image area of ​​an existing neuron. While training the network, the CNN automatically learns the sums through the filter layer based on how the user performs.

In particular, the basic structure of CNN usually includes 3 main parts:

  • Local receptive field (local field): This layer is responsible for filtering data, image information and selecting the image areas with the highest total value.
  • Shared weights and bias (shared weight): This layer helps to reduce the maximum number of parameters that have the main use of this factor in the CNN network. In each convolution there will be different feature maps and each feature has the ability to help detect some features in the image.
  • Pooling layer (composite layer): Pooling layer is the last layer and is used to simplify the output information. That is, after completing the calculation and scanning through the layers, go to the pooling layer to remove unnecessary information. From there, produce the results as desired by the user.

CNN is very widely used

Instructions on how to choose CNN parameters

To choose the most suitable parameter for CNN, you need to pay attention to the numbers: filter size, pooling size, number of convolution and training test.

  • Convolution layer: If this layer has a large number, your program will be even more advanced. Using layers with a large number of layers is likely to result in a significantly reduced impact. Usually after only 3 to 4 layers you will achieve the desired results.
  • Filter size: Usually, the filter sizes are 3×3 or 5×5.
  • Pooling size: For regular images, you need to use 2×2. otherwise, if the input has a larger image then you should use 4×4.
  • Train test: It is necessary to perform the train test many times, so that the best parameters can be obtained.

The Convolutional neural network algorithm gives the user a quality model. Although it is not a very simple algorithm by nature, it gives satisfactory results. However, this is a rather confusing algorithm and needs to go through a long exposure for users to be able to apply it correctly. Because, it is very difficult to know and understand CNN if you are a newcomer. Therefore, if you want to effectively apply CNN, you should learn and explore as well as add more knowledge for yourself. Hopefully, with the above sharing of ITNavi, you have understood What is CNN? as well as the structure of the CNN network.

How did you find the article?

Questions about What is CNN Network Structure Information?

Team ASIANA, in detail, Y Nhi has compiled the article based on available materials and knowledge from the Internet. Of course, we know that there are many questions and content that do not satisfy your requirements.

However, with the spirit of absorbing and improving, I always receive all compliments and criticisms from you & your readers for the article What is the Information about CNN Network Structure?

If you have any questions, what is CNN Network Structure Information? Please let us know, every eye or suggestion will help me improve more in the following articles.

Images About Information About Network Structure What is CNN?

Search keywords for the article #Things #Information #About #Structure #Network #CNN #What is #What

Find more information about What is CNN Network Structure Information? at WikiPedia

Please see more content about What is CNN Network Structure Information? from the Vietnamese Wikipedia website.◄

Join the Community At

💝 News source at: https://yt2byt.edu.vn

💝 See More Answers to Questions at: https://yt2byt.edu.vn/la-gi/

Bạn thấy bài viết Những Thông Tin Về Cấu Trúc Mạng CNN Là Gì? có giải quyết đươc vấn đề bạn tìm hiểu không?, nếu  không hãy comment góp ý thêm về Những Thông Tin Về Cấu Trúc Mạng CNN Là Gì? bên dưới để yt2byt.edu.vn có thể chỉnh sửa & cải thiện nội dung tốt hơn cho độc giả nhé! Cám ơn bạn đã ghé thăm Website Trường Cao đẳng Kỹ thuật Y tế II

Chuyên mục: Là gì?
#Những #Thông #Tin #Về #Cấu #Trúc #Mạng #CNN #Là #Gì

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button