News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
Understand the difference between AI, deep learning, and machine learning
Understand the difference between AI, deep learning, and machine learning
AI (Artificial Intelligence) is the future, and it is a science fiction novel, which is part of our daily life. All conclusions are correct, just to see what AI you are talking about.
For example, when Google's DeepMind developed the AlphaGo program to defeat Lee Se-dol, a Korean professional go expert, the media used AI, machine learning, and deep learning to describe the success of DeepMind. AlphaGo defeated Lee Se-dol, and all three technologies have made great achievements, but they are not the same thing.
To find out their relationship, the most intuitive way of expression is concentricity. The first is to present ideas, then machine learning. When machine learning is prosperous, it presents deep learning. Today's AI is driven by deep learning. .
A picture to understand the difference between AI, machine learning and deep learning
From decline to prosperity
In 1956, at the Dartmouth Conferences, computer scientists first proposed the term "AI", and AI was born. In the following days, AI became the "dream object" of the laboratory. After several decades, people's insights into AI have changed from time to time. Sometimes they think that AI is a sign, which is the key to human culture in the future. Sometimes it is just a technical waste, just a rash concept, too ambitious, and must fail. To be frank, until 2012, AI still has both of these characteristics.
In the past few years, AI has been in full swing, and it has been developing rapidly since 2015. The reason for the rapid development is mainly due to the general improvement of GPU, which makes parallel processing faster, cheaper and more powerful. Another reason is that the practice storage capacity is infinitely expanded, and data is generated in a wide range, such as pictures, texts, sales, and map data information.
AI: Let the machine show human intelligence
Back in the summer of 1956, at the time of the meeting, the illusion of the AI pioneer was to build a complex machine (the computer driver that was just presented) and then let the machine present the characteristics of human intelligence.
This concept is what we call "General AI", which is to create a superb machine that has all the human perceptions, and can also transcend human perception. It can be considered like a human being. . We often see such machines in movies, such as C-3PO, Terminator.
Another concept is "Narrow AI." Simply put, "weak artificial intelligence" can perform some detailed tasks like humans, and it is possible to do better than humans. For example, Pinterest serves to classify pictures with AI, and Facebook uses AI to identify faces. This is "weak labor." intelligent".
The above examples are examples of the use of "weak artificial intelligence", which have shown some of the characteristics of human intelligence. How is it done? Where does this intelligence come from? With the problem we deepen our understanding, we will come to the next circle, it is machine learning.
Machine Learning: A Way to Arrive AI Purpose
Generally speaking, machine learning uses algorithms to truly parse data, learn from time to time, and then make judgments and predictions about what is happening in the world. At this point, the researcher will not write the software by hand, affirm the special instruction set, and then let the program complete the special task. On the contrary, the researcher will use a lot of data and algorithms to "exercise" the machine and let the machine learn how to perform the task.
The concept of machine learning was proposed by early AI researchers. In the past few years, machine learning has presented many algorithmic approaches, including decision tree learning, categorization logic programming, clustering, reinforcement learning, Bayesian. Network, etc. As everyone knows, no one really reaches the ultimate goal of "strong worker intelligence". With the early machine learning approach, we have not even reached the goal of "weak artificial intelligence."
In the past many years, the best application case for machine learning has been "computer vision." To complete computer vision, researchers still need to manually write a lot of code to complete the task. The researcher manually writes the classifier, such as the edge detection filter, as long as the program can confirm where the object starts and where it is finished; the shape detection can confirm whether the object has 8 edges; the classifier can recognize the character "S-T-O-P". After manually writing the grouper, the researcher can develop an algorithm to identify a meaningful image, and then learn to discriminate, affirming that it is not a stop sign.
This method works, but it's not very good. If it is in a foggy day, when the visibility of the sign is relatively low, or if a tree blocks a part of the sign, its identification will fall. Until recently, computer vision and image detection technology were far removed from human talent because it was too error-prone.
Deep learning: the technology to complete machine learning
"Artificial Neural Networks" is another algorithmic approach that was proposed by early machine learning experts and has existed for decades. The idea behind Neural Networks stems from our understanding of the human brain—the interaction of neurons. There are also differences between the two. The neurons of the human brain are connected at specific physical intervals. The artificial neural network has independent layers, connections, and data transmission directions.
For example, you might take a picture, cut it into many pieces, and then implant it into the first layer of the neural network. The first layer of independent neurons transmits the data to the second layer, and the second layer of neurons also has its own task, which continues until the last layer and produces the final result.
Each neuron stops weighing the input information, affirming the weight, and figuring out its relationship with the task being performed, such as how accurate or how incorrect it is. The final result is determined by all weights. Taking the stop sign as an example, we will cut the stop sign image and let the neurons detect it, such as its octagonal shape, red, distinctive characters, traffic sign size, gestures, and so on.
The task of the neural network is to give a conclusion: whether it is a stop sign. The neural network gives a "probability vector" that relies on well-founded guesses and weights. In this case, the system has 86% confidence that the picture is a stop sign, 7% of self-confidence is sure it is a speed limit sign, 5% of self-confidence is sure it is a kite stuck in the tree, and so on. Then the network architecture will inform the neural network whether its discrimination is correct.
Even a simple thing is very advanced. Not long ago, the AI discussion community was still escaping the neural network. There was a neural network in the early days of AI, but it did not constitute a few "intelligence." The problem is that even if it is only a fundamental neural network, its request for computational complexity is high and cannot be a practical approach. Even so, there are still a handful of research teams going forward. For example, the team directed by Geoffrey Hinton of the University of Toronto, they put the algorithm parallel into the supercomputer, to verify their concept, until the GPU is generally adopted, we really see hope.
Going back to the example of identifying the abort sign, if we stop exercising on the network, exercise the network with a lot of wrong answers, and adjust the network, the result will be better. What the researchers need to do is exercise. They need to collect tens of thousands and even millions of pictures until the weight of the artificial neurons is highly accurate, so that every judgment is correct - whether it is foggy or foggy. The sun is bright and the rain is not affected. At this time, the neural network can "teach" himself, and find out exactly what the stop sign is; it can also identify Facebook's face image and recognize the cat.
The break is: make the neural network become very ambitious, increase the number of layers and neurons from time to time, let the system run a lot of data, exercise it.
Today, in some scenarios, machines that have been trained in deep learning techniques are better at recognizing images than humans, such as recognizing cats, identifying cancer cell features in the blood, and identifying tumors in MRI scans. Google AlphaGo learns Go, and he and I play Go and learn from time to time.
With the deep learning AI, the future will be bright
With deep learning, machine learning has many practical applications, and it has expanded the overall scope of AI. Deep learning splits the task, making machine assistance for each type possible. Driverless cars, better preventive treatments, and better movie referrals have either been presented or presented. AI is both today and the future. With the help of deep learning, perhaps one day AI will reach the extent of science fiction depiction, which is what we have been waiting for for a long time. You will have your own C-3PO and have your own terminator.
For example, when Google's DeepMind developed the AlphaGo program to defeat Lee Se-dol, a Korean professional go expert, the media used AI, machine learning, and deep learning to describe the success of DeepMind. AlphaGo defeated Lee Se-dol, and all three technologies have made great achievements, but they are not the same thing.
To find out their relationship, the most intuitive way of expression is concentricity. The first is to present ideas, then machine learning. When machine learning is prosperous, it presents deep learning. Today's AI is driven by deep learning. .
A picture to understand the difference between AI, machine learning and deep learning
From decline to prosperity
In 1956, at the Dartmouth Conferences, computer scientists first proposed the term "AI", and AI was born. In the following days, AI became the "dream object" of the laboratory. After several decades, people's insights into AI have changed from time to time. Sometimes they think that AI is a sign, which is the key to human culture in the future. Sometimes it is just a technical waste, just a rash concept, too ambitious, and must fail. To be frank, until 2012, AI still has both of these characteristics.
In the past few years, AI has been in full swing, and it has been developing rapidly since 2015. The reason for the rapid development is mainly due to the general improvement of GPU, which makes parallel processing faster, cheaper and more powerful. Another reason is that the practice storage capacity is infinitely expanded, and data is generated in a wide range, such as pictures, texts, sales, and map data information.
AI: Let the machine show human intelligence
Back in the summer of 1956, at the time of the meeting, the illusion of the AI pioneer was to build a complex machine (the computer driver that was just presented) and then let the machine present the characteristics of human intelligence.
This concept is what we call "General AI", which is to create a superb machine that has all the human perceptions, and can also transcend human perception. It can be considered like a human being. . We often see such machines in movies, such as C-3PO, Terminator.
Another concept is "Narrow AI." Simply put, "weak artificial intelligence" can perform some detailed tasks like humans, and it is possible to do better than humans. For example, Pinterest serves to classify pictures with AI, and Facebook uses AI to identify faces. This is "weak labor." intelligent".
The above examples are examples of the use of "weak artificial intelligence", which have shown some of the characteristics of human intelligence. How is it done? Where does this intelligence come from? With the problem we deepen our understanding, we will come to the next circle, it is machine learning.
Machine Learning: A Way to Arrive AI Purpose
Generally speaking, machine learning uses algorithms to truly parse data, learn from time to time, and then make judgments and predictions about what is happening in the world. At this point, the researcher will not write the software by hand, affirm the special instruction set, and then let the program complete the special task. On the contrary, the researcher will use a lot of data and algorithms to "exercise" the machine and let the machine learn how to perform the task.
The concept of machine learning was proposed by early AI researchers. In the past few years, machine learning has presented many algorithmic approaches, including decision tree learning, categorization logic programming, clustering, reinforcement learning, Bayesian. Network, etc. As everyone knows, no one really reaches the ultimate goal of "strong worker intelligence". With the early machine learning approach, we have not even reached the goal of "weak artificial intelligence."
In the past many years, the best application case for machine learning has been "computer vision." To complete computer vision, researchers still need to manually write a lot of code to complete the task. The researcher manually writes the classifier, such as the edge detection filter, as long as the program can confirm where the object starts and where it is finished; the shape detection can confirm whether the object has 8 edges; the classifier can recognize the character "S-T-O-P". After manually writing the grouper, the researcher can develop an algorithm to identify a meaningful image, and then learn to discriminate, affirming that it is not a stop sign.
This method works, but it's not very good. If it is in a foggy day, when the visibility of the sign is relatively low, or if a tree blocks a part of the sign, its identification will fall. Until recently, computer vision and image detection technology were far removed from human talent because it was too error-prone.
Deep learning: the technology to complete machine learning
"Artificial Neural Networks" is another algorithmic approach that was proposed by early machine learning experts and has existed for decades. The idea behind Neural Networks stems from our understanding of the human brain—the interaction of neurons. There are also differences between the two. The neurons of the human brain are connected at specific physical intervals. The artificial neural network has independent layers, connections, and data transmission directions.
For example, you might take a picture, cut it into many pieces, and then implant it into the first layer of the neural network. The first layer of independent neurons transmits the data to the second layer, and the second layer of neurons also has its own task, which continues until the last layer and produces the final result.
Each neuron stops weighing the input information, affirming the weight, and figuring out its relationship with the task being performed, such as how accurate or how incorrect it is. The final result is determined by all weights. Taking the stop sign as an example, we will cut the stop sign image and let the neurons detect it, such as its octagonal shape, red, distinctive characters, traffic sign size, gestures, and so on.
The task of the neural network is to give a conclusion: whether it is a stop sign. The neural network gives a "probability vector" that relies on well-founded guesses and weights. In this case, the system has 86% confidence that the picture is a stop sign, 7% of self-confidence is sure it is a speed limit sign, 5% of self-confidence is sure it is a kite stuck in the tree, and so on. Then the network architecture will inform the neural network whether its discrimination is correct.
Even a simple thing is very advanced. Not long ago, the AI discussion community was still escaping the neural network. There was a neural network in the early days of AI, but it did not constitute a few "intelligence." The problem is that even if it is only a fundamental neural network, its request for computational complexity is high and cannot be a practical approach. Even so, there are still a handful of research teams going forward. For example, the team directed by Geoffrey Hinton of the University of Toronto, they put the algorithm parallel into the supercomputer, to verify their concept, until the GPU is generally adopted, we really see hope.
Going back to the example of identifying the abort sign, if we stop exercising on the network, exercise the network with a lot of wrong answers, and adjust the network, the result will be better. What the researchers need to do is exercise. They need to collect tens of thousands and even millions of pictures until the weight of the artificial neurons is highly accurate, so that every judgment is correct - whether it is foggy or foggy. The sun is bright and the rain is not affected. At this time, the neural network can "teach" himself, and find out exactly what the stop sign is; it can also identify Facebook's face image and recognize the cat.
The break is: make the neural network become very ambitious, increase the number of layers and neurons from time to time, let the system run a lot of data, exercise it.
Today, in some scenarios, machines that have been trained in deep learning techniques are better at recognizing images than humans, such as recognizing cats, identifying cancer cell features in the blood, and identifying tumors in MRI scans. Google AlphaGo learns Go, and he and I play Go and learn from time to time.
With the deep learning AI, the future will be bright
With deep learning, machine learning has many practical applications, and it has expanded the overall scope of AI. Deep learning splits the task, making machine assistance for each type possible. Driverless cars, better preventive treatments, and better movie referrals have either been presented or presented. AI is both today and the future. With the help of deep learning, perhaps one day AI will reach the extent of science fiction depiction, which is what we have been waiting for for a long time. You will have your own C-3PO and have your own terminator.