Differences between artificial intelligence, machine learning and deep learning
Artificial intelligence, machine learning and deep learning are already an integral part of many enterprises. Often these terms are used interchangeably.
Artificial intelligence moves in enormous strides — from advances in unmanned vehicles and the ability to beat people in games like poker and go to automated customer service. Artificial intelligence is an advanced technology that is ready to revolutionize business.
Often the terms artificial intelligence, machine learning and deep learning are used haphazardly interchangeably, but in fact there are differences between them. What is the difference between these terms will be described below.
Artificial intelligence is a broad concept that refers to advanced machine intelligence. In 1956, at a conference on artificial intelligence in Dartmouth, this technology was described as follows: “Every aspect of training or any other feature of intelligence can in principle be described so accurately that a machine can imitate them.”
')
Artificial intelligence can relate to anything - from computer programs for playing chess to speech recognition systems, such as, for example, Amazon Alexa voice assistant, capable of perceiving speech and answering questions. In general, artificial intelligence systems can be divided into three groups: limited artificial intelligence (Narrow AI), general artificial intelligence (AGI), and supramental artificial intelligence.
IBM’s Deep Blue program, which beat Garry Kasparov in chess in 1996, or Google’s DeepMind program AlphaGo , which beat world champion Guo Li Sedol in 2016, are examples of limited artificial intelligence capable of solving one specific problem. This is its main difference from general artificial intelligence (AGI), which is on the same level with human intelligence and can perform many different tasks.
Superintelligent artificial intelligence is a step higher than human. Nick Bostrom describes it as follows: it is “an intellect that is much smarter than the best human brain, in almost all areas, including scientific work, general wisdom and social skills.” In other words, this is when machines become much smarter than us.
Machine learning is one of the areas of artificial intelligence. The basic principle is that machines receive data and are “trained” on them. Currently it is the most promising business tool based on artificial intelligence. Machine learning systems allow you to quickly apply the knowledge gained during training on large data sets, which allows them to succeed in such tasks as face recognition, speech recognition, object recognition, translation, and many others. Unlike programs with manually coded instructions for performing specific tasks, machine learning allows the system to learn how to recognize patterns and make predictions on its own.
While both programs — Deep Blue and DeepMind — are examples of using artificial intelligence, Deep Blue was built on a pre-programmed set of rules, so it has nothing to do with machine learning. DeepMind, on the other hand, is an example of machine learning: the program beat the world Go champion, teaching herself on a large set of move data made by experienced players.
Is your business interested in integrating machine learning into your strategy? Amazon, Baidu, Google, IBM, Microsoft, and others are already offering machine learning platforms that businesses can use.
Deep learning is a subset of machine learning. It uses some machine learning techniques to solve real-world problems using neural networks that can mimic human decision making. Deep learning can be costly and requires huge amounts of data for training. This is due to the fact that there are a huge number of parameters that must be configured for learning algorithms in order to avoid false positives. For example, the deep learning algorithm may be instructed to “find out” what a cat looks like. To produce training, you need a huge number of images in order to learn to distinguish the smallest details that allow you to distinguish a cat from, say, a cheetah or a panther, or a fox.
As mentioned above, a major victory was achieved by artificial intelligence in March 2016, when the AlphaGo DeepMind program beat world champion Guo Lee Sedol in 4 out of 5 games using deep learning. As Google explained, the deep learning system worked by combining " the Monte Carlo method for searching the tree with deep neural networks that were trained with a teacher at the professional games and reinforcement training at the games with themselves."
Deep learning also has business applications. You can take a huge amount of data - millions of images, and use them to identify certain characteristics. Text search, fraud detection, spam detection, handwriting recognition, image search, speech recognition, translation - all these tasks can be accomplished with in-depth training. For example, in Google, deep learning networks have replaced many "rule-based systems that require manual work."
It is worth noting that deep learning can be very “biased”. For example, when Google’s face recognition system was initially deployed, it tagged many black faces as gorillas. “This is an example of what will happen if you do not have African-American people in your training suite,” said Anu Tewary, Chief Data Specialist at Mint at Intuit. “If you don’t have African Americans working on the system, if you don’t have African Americans testing the system, then when your system encounters African American people, it won’t know how to behave.”
There is an opinion that the subject of deep learning is greatly inflated . The Sundown AI system, for example, provides automated customer interactions using a combination of machine learning and a policy graph of algorithms without the use of deep learning.
The original article is “ Understanding the differences between AI, machine learning, and deep learning .”
Artificial intelligence moves in enormous strides — from advances in unmanned vehicles and the ability to beat people in games like poker and go to automated customer service. Artificial intelligence is an advanced technology that is ready to revolutionize business.
Often the terms artificial intelligence, machine learning and deep learning are used haphazardly interchangeably, but in fact there are differences between them. What is the difference between these terms will be described below.
Artificial Intelligence (AI)
Artificial intelligence is a broad concept that refers to advanced machine intelligence. In 1956, at a conference on artificial intelligence in Dartmouth, this technology was described as follows: “Every aspect of training or any other feature of intelligence can in principle be described so accurately that a machine can imitate them.”
')
Artificial intelligence can relate to anything - from computer programs for playing chess to speech recognition systems, such as, for example, Amazon Alexa voice assistant, capable of perceiving speech and answering questions. In general, artificial intelligence systems can be divided into three groups: limited artificial intelligence (Narrow AI), general artificial intelligence (AGI), and supramental artificial intelligence.
IBM’s Deep Blue program, which beat Garry Kasparov in chess in 1996, or Google’s DeepMind program AlphaGo , which beat world champion Guo Li Sedol in 2016, are examples of limited artificial intelligence capable of solving one specific problem. This is its main difference from general artificial intelligence (AGI), which is on the same level with human intelligence and can perform many different tasks.
Superintelligent artificial intelligence is a step higher than human. Nick Bostrom describes it as follows: it is “an intellect that is much smarter than the best human brain, in almost all areas, including scientific work, general wisdom and social skills.” In other words, this is when machines become much smarter than us.
Machine learning
Machine learning is one of the areas of artificial intelligence. The basic principle is that machines receive data and are “trained” on them. Currently it is the most promising business tool based on artificial intelligence. Machine learning systems allow you to quickly apply the knowledge gained during training on large data sets, which allows them to succeed in such tasks as face recognition, speech recognition, object recognition, translation, and many others. Unlike programs with manually coded instructions for performing specific tasks, machine learning allows the system to learn how to recognize patterns and make predictions on its own.
While both programs — Deep Blue and DeepMind — are examples of using artificial intelligence, Deep Blue was built on a pre-programmed set of rules, so it has nothing to do with machine learning. DeepMind, on the other hand, is an example of machine learning: the program beat the world Go champion, teaching herself on a large set of move data made by experienced players.
Is your business interested in integrating machine learning into your strategy? Amazon, Baidu, Google, IBM, Microsoft, and others are already offering machine learning platforms that businesses can use.
Deep learning
Deep learning is a subset of machine learning. It uses some machine learning techniques to solve real-world problems using neural networks that can mimic human decision making. Deep learning can be costly and requires huge amounts of data for training. This is due to the fact that there are a huge number of parameters that must be configured for learning algorithms in order to avoid false positives. For example, the deep learning algorithm may be instructed to “find out” what a cat looks like. To produce training, you need a huge number of images in order to learn to distinguish the smallest details that allow you to distinguish a cat from, say, a cheetah or a panther, or a fox.
As mentioned above, a major victory was achieved by artificial intelligence in March 2016, when the AlphaGo DeepMind program beat world champion Guo Lee Sedol in 4 out of 5 games using deep learning. As Google explained, the deep learning system worked by combining " the Monte Carlo method for searching the tree with deep neural networks that were trained with a teacher at the professional games and reinforcement training at the games with themselves."
Deep learning also has business applications. You can take a huge amount of data - millions of images, and use them to identify certain characteristics. Text search, fraud detection, spam detection, handwriting recognition, image search, speech recognition, translation - all these tasks can be accomplished with in-depth training. For example, in Google, deep learning networks have replaced many "rule-based systems that require manual work."
It is worth noting that deep learning can be very “biased”. For example, when Google’s face recognition system was initially deployed, it tagged many black faces as gorillas. “This is an example of what will happen if you do not have African-American people in your training suite,” said Anu Tewary, Chief Data Specialist at Mint at Intuit. “If you don’t have African Americans working on the system, if you don’t have African Americans testing the system, then when your system encounters African American people, it won’t know how to behave.”
There is an opinion that the subject of deep learning is greatly inflated . The Sundown AI system, for example, provides automated customer interactions using a combination of machine learning and a policy graph of algorithms without the use of deep learning.
The original article is “ Understanding the differences between AI, machine learning, and deep learning .”
Source: https://habr.com/ru/post/401857/
All Articles