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
Machine learning and artificial intelligence
Machine learning and artificial intelligence
In the past few years, machine learning and artificial intelligence have made great progress in accuracy. However, regulated industries, such as banks, are still indecisive, often giving priority to the accuracy and efficiency of regulatory compliance and algorithm interpretation. Some companies think this technology is unbelievable or risky.
.
During the financial crisis of 2008, the banking industry realized that their machine learning algorithms were based on defective assumptions. Therefore, regulators of financial system need additional control measures and introduce regulatory requests for banks and insurance companies to stop "formal risk" management.
Banks must also prove that they understand the models they use, so unfortunately, we can understand that they intentionally limit their technological complexity and adopt a generalized linear model which is simpler and more explanatory.
If you want to build trust in machine learning, you can try to look at it like a person and ask it the same question.
In order to trust the initiative provided by AI and machine learning, the needs of enterprises from all industries are trying to better understand it. Data scientists and Dr. should not be the only thing you can clearly explain the machine learning model, because as AI theorist Eliezer Yudkowsky said: "so far, the biggest risk of artificial intelligence is that people think they prematurely understand the technology.
A human approach to trust demand
When asked about the data scientists machine learning model is how to make a resolution of the time, they tend to use complex mathematical equations to solve, so that outsiders do not know how dumb as a wooden chicken, can trust this model. Will it be more effective to deal with machine learning decisions in the same way as human decision-making? As Udacity, the founder of Udacity (Sebastian Thrun) once said, "artificial intelligence is simply a humanities. This practice is an attempt to understand human intelligence and human cognition. "
So do not use complex mathematical equations to confirm how the loan officer makes decisions, but simply ask, "what is the most important information on the loan application form for your decision? Or, "what is the value of the risk up and down, and how do you decide to accept or reject some specific loan applications?"
The same human method can be used to confirm how the algorithm makes similar resolutions. For example, by using machine learning technology called characteristic influence, we can affirm the balance of circulating efficiency, and the applicant's income and loan purpose are the three most important information of the loan officer algorithm.
Through the use of reason is called code, people can see the most important factor to measure detailed materials for each loan applicant in, and after the application is called local dependence on technology, the algorithm will be able to see higher income loan application risk rating is low.
The value of objectivity, extensibility and predictability
By analyzing how machines make decisions like human beings, we can make human beings better understand artificial intelligence and machine learning. In addition, human beings can realize the trust of AI and machine learning by recognizing the common ability of technology, including:
Deal with the problem of credibility and data outliers: the pass through model usually requires the assumption of how data is created, the process behind data and the credibility of data. However, machine learning is using highly sensitive algorithms to eliminate these restrictive assumptions. These algorithms will not give more credibility than it deserves.
Support modern computers and mass data sets: unlike handmade processes, machine learning does not assume that the world is full of lines. Instead, it will automatically adjust the equation to find out the best form, and test which algorithms and forms are most suitable for independent textual data, rather than just testing the data that they exercise.
The application value of the future: lack of advanced machine learning is not a request for several hours of data clearing, but to build a blueprint for the optimization of specific algorithms, automatic detection of missing values, determine which algorithms are not applicable for replacing missing values, the optimal value of missing values, and the use of the missing value exists to predict different results.
Do not doubt the initiative of AI or machine learning, let us interrogate us to ask for the same reasoning problem of human beings to better understand them. Let us recognize the objective ability of technology to reduce the credibility of data exceptions, and to provide extensible agility for today's massive data.
Perhaps most importantly, let us confess the ability of AI and machine learning to better predict future results by using short information. Because technology is strong enough to be vigilant and formal supervision, but if we can establish a correct understanding and trust degree, consumers and enterprises will only benefit.
.
During the financial crisis of 2008, the banking industry realized that their machine learning algorithms were based on defective assumptions. Therefore, regulators of financial system need additional control measures and introduce regulatory requests for banks and insurance companies to stop "formal risk" management.
Banks must also prove that they understand the models they use, so unfortunately, we can understand that they intentionally limit their technological complexity and adopt a generalized linear model which is simpler and more explanatory.
If you want to build trust in machine learning, you can try to look at it like a person and ask it the same question.
In order to trust the initiative provided by AI and machine learning, the needs of enterprises from all industries are trying to better understand it. Data scientists and Dr. should not be the only thing you can clearly explain the machine learning model, because as AI theorist Eliezer Yudkowsky said: "so far, the biggest risk of artificial intelligence is that people think they prematurely understand the technology.
A human approach to trust demand
When asked about the data scientists machine learning model is how to make a resolution of the time, they tend to use complex mathematical equations to solve, so that outsiders do not know how dumb as a wooden chicken, can trust this model. Will it be more effective to deal with machine learning decisions in the same way as human decision-making? As Udacity, the founder of Udacity (Sebastian Thrun) once said, "artificial intelligence is simply a humanities. This practice is an attempt to understand human intelligence and human cognition. "
So do not use complex mathematical equations to confirm how the loan officer makes decisions, but simply ask, "what is the most important information on the loan application form for your decision? Or, "what is the value of the risk up and down, and how do you decide to accept or reject some specific loan applications?"
The same human method can be used to confirm how the algorithm makes similar resolutions. For example, by using machine learning technology called characteristic influence, we can affirm the balance of circulating efficiency, and the applicant's income and loan purpose are the three most important information of the loan officer algorithm.
Through the use of reason is called code, people can see the most important factor to measure detailed materials for each loan applicant in, and after the application is called local dependence on technology, the algorithm will be able to see higher income loan application risk rating is low.
The value of objectivity, extensibility and predictability
By analyzing how machines make decisions like human beings, we can make human beings better understand artificial intelligence and machine learning. In addition, human beings can realize the trust of AI and machine learning by recognizing the common ability of technology, including:
Deal with the problem of credibility and data outliers: the pass through model usually requires the assumption of how data is created, the process behind data and the credibility of data. However, machine learning is using highly sensitive algorithms to eliminate these restrictive assumptions. These algorithms will not give more credibility than it deserves.
Support modern computers and mass data sets: unlike handmade processes, machine learning does not assume that the world is full of lines. Instead, it will automatically adjust the equation to find out the best form, and test which algorithms and forms are most suitable for independent textual data, rather than just testing the data that they exercise.
The application value of the future: lack of advanced machine learning is not a request for several hours of data clearing, but to build a blueprint for the optimization of specific algorithms, automatic detection of missing values, determine which algorithms are not applicable for replacing missing values, the optimal value of missing values, and the use of the missing value exists to predict different results.
Do not doubt the initiative of AI or machine learning, let us interrogate us to ask for the same reasoning problem of human beings to better understand them. Let us recognize the objective ability of technology to reduce the credibility of data exceptions, and to provide extensible agility for today's massive data.
Perhaps most importantly, let us confess the ability of AI and machine learning to better predict future results by using short information. Because technology is strong enough to be vigilant and formal supervision, but if we can establish a correct understanding and trust degree, consumers and enterprises will only benefit.