The history of human civilization is the history of manufacturing. Man invented many tools to improve efficiency. But, no matter what kind of tools, man is always the subject, and tools are only an extension of the human hands. People expect the machine to work for people and hope that the machine also has the same intelligence as people think. Figure 2-19 shows the move from abacus to a calculator to machine learning, from manual to automatic driving to driverless driving.
In computer science, in contrast to natural human intelligence, artificial intelligence (AI) (sometimes called machine intelligence) is the intelligence demonstrated by machines.
Figure 2-19 Humans continue to free their brains
The term “artificial intelligence” is commonly used to describe machines (or software systems running on computers) that mimic the “cognitive” functions associated with human thinking (e.g., “learning” and “problem-solving”) of a machine (or software system running on a computer). Thus, artificial intelligence refers to anything that can replace human intelligence and can be considered artificial intelligence. However, due to the continuous development of machine learning technology, AI has made significant breakthroughs in speech recognition and image recognition. Therefore, traditional recognition technology, such as OCR font recognition technology, is not considered artificial intelligence.
Are Artificial Intelligence, Machine Learning, and Deep Learning the same thing? How can we tell the difference and relationship between them?
In the history of the development of artificial intelligence, at the beginning stage, there was always an attempt to give intelligence to machines by humans, including giving them the logical ability and expert knowledge bases, so that the machines would have intelligence. But it was soon found that it is difficult to make the machine have real intelligence. It is hard to reason with mathematical logic and knowledge base, such as image recognition and speech recognition, but it is easy to do it with a child.
Can we let the machine learn by itself? Just like a toddler learns language, recognizes cars, etc., it can have intelligence by learning through known things, repeating the learning repeatedly, extracting features, and then building recognition models based on the features. This is called machine learning.
Figure 2-20 explains the relationship between artificial intelligence, machine learning, and deep learning. Artificial intelligence includes strong logical reasoning and cognitive-based reasoning, and machine learning is a branch of artificial intelligence. Among machine learning, deep learning algorithms have become an important branch of machine learning because of the remarkable graphics and speech recognition results.
Figure 2-20 Relationship between artificial intelligence, machine learning, and deep learning
Traditional machine learning algorithms, often, human beings extract the features to be recognized to take and then based on different algorithms, such as convolutional algorithms, to build models so that inference and judgment can be achieved. How to extract the features of things is a very difficult problem. What are the features of things, need to spend a lot of time summarizing and organizing, and refine the features? For example, language recognition, even if it is a kind of problem, there are various dialects in different places. How to extract the features?
Can we let the machine extract the features itself? Thus came the deep learning algorithm, which uses a multilayer neural network algorithm to integrate feature extraction and model training, creation, and testing, which greatly simplifies the process and involves less human involvement.
Artificial intelligence, big data, cloud computing, the three relationships are: artificial intelligence needs big data to extract models, and big data needs cloud computing support. With the combination of the three, human beings enter the era of artificial intelligence.
Artificial intelligence is relevant to human intellectual tasks. Modern AI technologies are ubiquitous, and compelling examples of AI include self-driving cars (e.g., drones and self-driving cars), medical diagnostics, creating art (e.g., poetry), proving mathematical theorems, playing games (e.g., chess or Go), search engines (e.g., Google search), online assistants (e.g., Siri), image recognition in photographs, spam filtering, predicting flight delays, predicting judicial decisions, targeting online ads, etc. Applications of artificial intelligence in agriculture are called smart farms, and in industry, smart factories. These specific applications include News, healthcare, automotive, finance, cybersecurity, government, legal-related, video games, military, service industry, auditing, advertising, art, etc.
In the field of smart manufacturing, applications of AI include:
- Intelligent models of products can be customized, the process changes automatically with the design, and data can be generated automatically.
- Part identification: visually identifies the model and specification of the part and generates corresponding data and actions based on the identification result.
- Self-driving: Apply self-driving technology to realize unmanned logistics in factories.
- Automatic storage: Using vision technology to achieve automated outbound and inbound storage.
- Intelligent quality control: use AI technology to measure the appearance of products, various parameters.
- Intelligent grinding: human-like realization of unmanned machine tooling.
- Intelligent painting: unmanned painting like a human, absorbing the best human painting techniques.
- Fault judgment: The machine operation model is established by sound, and other parameters, so abnormal judgment can be made, and losses caused by downtime can be avoided.
Of course, there will be more AI technologies for smart manufacturing. However, the ultimate goal of smart factories is to be unmanned, and AI can make smart factories keep approaching this goal.