Chen Jidong, biometric identification Director at Ants Financial Services about "Face Sweeping&

The Alipay "Face Sweeping" technology has now more than 150 million users worldwide and made it to the 2017 top 10 Breakthrough Technologies of the MIT Review . Here's an interview of the man behind the technology.

When speaking about entrepreneurship and innovation, it seems very difficult to get rid of ‘foreign’, ‘Silicon Valley’ or other slightly stereotyped terms. Even if today the trend has been reversed, many outstanding Chinese enterprises and technological innovations have been completely ignored due to the lack of appropriate media coverage and display platforms.

In the ‘MIT Science and Technology Review’, the ‘Face Sweeping Payment’ technology has made the world's top 10 breakthrough technologies selection in 2017, and caught everyone’s attention because its main representatives are all from China. In fact, when it comes to ‘face sweeping payment’ related applications - which are based on face recognition, China is now in the forefront of the world, far ahead of the United States and other countries.

In order to further analyze the technical details of the ‘face sweeping’ technology, and to understand the story behind China’s sudden emergence in the field of face recognition technology, we interviewed a senior expert in global identification platforms, the director of biometric identification technology at the Ants Financial Services, Chen Jidong.

‘Face Sweeping Payment’ has made the list of top 10 breakthrough technologies according to ‘MIT Science and Technology Review’. One representative of this technology is Alibaba. What can you say about this list?

To be precise, the company mentioned in the list is Ants Financial Services. Of course, Ants Financial Services is an affiliated company of Alibaba Group.

The descriptions in the top 10 list is relatively accurate. The breakthrough technology on the list is ‘Face Sweeping Payment’, not face recognition. The definition of the breakthrough is ‘to bring people high-quality technology solutions.’ So, it is not only about the technology itself, but also about how to apply technology in order to change people's lives.

Emphasizing on the application of technology is also the positioning ofAnts Financial Services. In addition to the technology itself, we payed more attention to how the application of technology can bring equal financial services to the world. Face recognition is a common technology that can be used in financial payments, offline security, entertainment, social contact and many other areas.

But in our view, using face recognition for financial services such as payment is greater than other uses, because identification or authentification are the basis of all financial services, and the online authentication based on face recognition played a very important role in common digital financial services.

At the same time, being different from the general direction of face recognition that lays in social contact and entertainment, ‘Face Sweeping Payment’ needs a financial level of accuracy and security.

In fact, the ‘Face Sweeping Payment’ technology is developed by the Ants Financial Services and Face ++. On the basis of the face recognition core alignment algorithm, Ants Financial Services has developed a patented technology of detecting living bodies, combined with its core technology, the risk controlling and anti-attack security strategy and other multiple dimensions of technologies. We are now able to improve accuracy and security on a financial level.

Chen Jidong, Ant Financial Services Group security intelligence director.

What is the size of Alipay 2C users? How is the development of ‘Face Sweeping Payment’ application?

Alipay currently has more than 450 million real-name users, within which more than 150 million users have used the ‘Face Sweeping Payment’ to log in their Alipay account, to retrieve passwords and to authenticate in high-risk transactions. Law requires that payment systems must adopt multi-channel authentification methods; nowadays binding credit cards with bank accounts and ‘Face Sweeping’ authentication are the primary methods. But it is clear that ‘Face Sweeping’ is a more convenient and safer choice.

In the development of Internet finance or Fintech, the requirement of real names and anonymous features are in fact contradictory. So the key question that needs to be answered is : ‘how do you prove that you are actually you’, and it is also a problem for the online financial services of traditional banks. In general, our research on face recognition technology is driven by pragmatic requirements.

From the 10 years Alipay experience we found out that identity and authentication are not only the foundation and the core of the payment system, but also that of all financial businesses. After all, in the context of ‘Internet plus’, everything is online. First of all we must identify the user, then provide our user with more secure and more convenient services.

The existing user authentication method is divided into the following categories:

1) what you have: phone, U shield;

2) what you know: password, security question;

3) the only thing you have: personal biology.

From the perspective of security and convenience, passwords and SMS authentication codes and other commonly used authentication methods hardly satisfy these two needs, and personal biology is the only promising approach left which can be relied on to prove that ‘you are actually you’. This is exactly the pragmatic ground for developing ‘Face Sweeping Payment’, which represents the applications based on face recognition in the financial field.

Overall, the ‘Face Sweeping Payment’ was born in the unique financial business of Alipay. This breakthrough technology is developed through collaborative research between the Ant Financial Services and Face ++, and its solution consists of the face matching algorithm provided by Face ++, the living body identification algorithm and the whole risk control & anti-attack technology system provided by Ant Financial Services.

What is the process of development since it was online?

Our product was officially launched in July, 2015. Before this we were doing all sorts of small-scale testing, product optimization and upgrades.

We found out that the real circumstances are very complicated. The users might be indoors or outdoors, in day time or or night time. They ‘sweep’ their faces under different lights, from different angles and with different postures, some of them are lying on their beds, and some of them are even wearing facial masks. Dealing with this level of complexity with ‘Face Sweeping Payment’ in order for normal users to have an easy access to the technology, has been a great challenge.

After more than a year of product optimization, we are now able to ensure that in a variety of complex environments the users can still enjoy a good ‘face sweeping’ experience.

What is the current user growth situation? Is it not in the acceleration phase?

In fact, for a long time, we have been doing a ‘grey-scale control’, we just wanted to optimize the algorithm to the best state of security and experience before launching a large-scale promotion. As for the users in Alipay, when they will encounter problems with the password input or receiving the SMS check code or any other problem, once they find a new technology that is secured and assured, they will take the initiative to use it. So, the growth of users is quite a natural thing.

When can the ‘Face Sweeping Payment’ have a large-scale expansion?

Now, we have opened full access to ‘Face Sweeping Payment’ and other functions to the Alipay users. They can log in their Alipay accounts without registering for ‘Face Sweeping Payment’. But compared to the volume, we are more concerned about the using experience and the security level of the product.

Alipay's current online ‘Face Sweeping’ function has been optimized, the upgrading of user experience and security level were simultaneously achieved, with a better result than the passwords and SMS check codes or any other traditional authentication method. At the same time, Ant Financial Services is also exploring the offline application of ‘Face Sweeping’. But since the offline circumstances are more complicated, it is mainly online for now.

Why did you produce ‘Face Sweeping Payment’ technology?

For the entire Alipay risk control system, authentification is the core issue. Because the core of the payment is to verify your identity, to know who you are, then confirm the operation. And face recognition has become a safe and convenient way of authentification. Moreover, not only in the payment, but also in the user’s log-in, real-name authentification, password retrieving, business review and many other aspects it will have a wide range of applications.

In addition, one of the most important applications for a financial service is the remote account of Internet banking. For example, Internet or online banks adopted face recognition technology in the process of opening an account. And, in the first place, the Ants’ face recognition technology is developed to solve the problem of online identification for online banks.

In fact, the biggest difference between online banks and physical banks is that there is no business outlets for the former, which means that the offline face-to-face process must be automated through technologies, meanwhile guaranteeing a high level of security and accuracy.

In contrast to offline comparison of documents and faces, we automate this process - through face recognition technology, document OCR (optical character recognition) technology and anti-counterfeiting technology, ensuring the cross-comparison among the ID cards and faces and the government authoritative identity library. With the judgments based on large multiple dimensional data, you can more accurately know that whether this person in front of you is holding the legal document. You can then open a new online account for the user.

Although the national policy on remote accounts has not yet been released, it is only possible to open a second-class account with limited services in this way. But it is testified by practices that our technology is feasible. It should be emphasized that face recognition plays a very important role in this remote account opening process.

How is the accuracy of ‘Face sweeping’ in reality ?

The most important thing face recognition has to overcome is the various angles, postures and light interference. To solve this problem, not only we have to solve the algorithm, but also problems related to the product, interaction, user-experience, environmental parameters and security policy and lots of other problems. This is system engineering, especially when the number of users reached the scale of 100 million.

First, we designed a quality inspection module to ensure that a sufficient quality of portrait was collected before the authentication was completed. Then we need a suitable guide page to teach users how to properly sweep their faces, in order to prevent the failure due to unsuitable postures which will affect the using experience. Through the intelligent algorithm, the app can automatically adjust the different parameters of the threshold according to the current environment and angle, so that the overall pass rate can be higher.

In addition to software, will hardware differences including the chips affect the ‘Face Sweeping Payment’?

Indeed, we are still developing a pure software system, but Alipay has 450 million users, we can not only serve a few high-end customers only. It does not match with our common-benefit philosophy .

Therefore, our algorithm must adapt to a variety of user-ends, especially to smart phones which are lower than 1 thousand yuan. To ensure that ‘Face Sweeping Payment’ can meet the needs of authentication, it is necessary to solve the problem of fragmentation in Android machines, that is to say, the application must be compatible with different cameras, different CPUs and other complex issues based on some intelligent algorithms.

Fortunately, compared to other biometric technologies, the hardware requirements for face recognition are not so high. It is more universal, compatible with any camera that can take photos. But of course the better the device, the higher the accuracy of recognition.

In business development part, what are the potential fields for ‘Face Sweeping Payment’ technology?

In addition to the financial system, face recognition is also applied in the online public services. For example, the online search for credit or taxation records requires people to authenticate their identities. At present, we have solved it with our unified identity authentication solution, the Zhima Authentification system.

Zhima Authentification system is now accessible to social security, public security, traffic police, tax and other city service agencies, such as Wuhan Public Security electronic ID card, Shenzhen People Club Bureau, Nanning Public Security, Beijing Local Tax and so on. These agencies can use the Alipay ‘Face Sweeping’ technology to deal with issues, greatly improving the degree of convenience. It used to be a very cumbersome thing for elder people to receive their social security, but now it has been dealt with.

What are the future applications for face recognition?

In fact, this is a very simple question of proving ‘you are you’ and in turn helping the organizations to know ‘who you are’. So this is actually related to all the services of Internet Plus.

Financial services need to know ‘who you are’ so they can understand their clients before they provide financial services such as transfers and loans. Credit system is the basic service of society, and identification or authentification are the basis of credit system. The core of all credit services is to know the personal credit rating, and of course, the premise is that to know ‘you you are’. Otherwise the evaluation of credit rating may turn out to be someone else's.

At present, many of our Internet services are slowly moving to real-name system, such as the purchase of train tickets, air tickets, cell phone numbers and so on. Previously, using 12306 to buy a train ticket would need complex graphical verification, but the application of face recognition can be a good solution to this problem.

So, authentification has become the infrastructure of online finance, and even that of the entire Internet Plus. The biometrics authentification method can do much better in proving that ‘you are you’, with a higher degree of convenience and security.

Many people are criticizing China's credit system for falling behind the United States, what do you think?

Authentification is the basis of credit, and from this perspective, I think China is now much ahead of the United States. For example, Chinese government has been building a centralized identity library almost 10 years ago, providing identity verification services for financial institutions and others. If the fingerprint library which is still in process ever reached a certain size, in fact, it can be used for authentification as well.

But the United States does not have a unified identity agency. If the US banks wish to KYC (Know Your Customer), they’ll have to wait for their customers to fill out the information so that they can find the data comparison of different states. Though there are many third parties continuing to integrate the data in order to provide related services, they can be improved significantly.

Original article

#facerecognition #technology #AntFinacial #Alipay