A Convenient System for Pension Insurance

Because of one-child policy, China is in an aging society now. Most of Chinese familes are 4-2-1 structure. One child, parents, 4 grandparents. In this situation, Pension Insurance is a very important topic. To grant the insurance to old people, the information collection is a basic process. The government should collect correct information to make sure every old people can benefit from the insurance and the money is not wrongly distributed.

Normally this procedure is done by visit and find out the living conditions of old people. But in this way, the working efficiency will be very low. Furthermore, later years it is important to ensure whether the person is still alive. With the development of artificial intelligent, you may think about using fingerprint collection. Fingerprint can be used as a survival certificate, it can make sure that each old people can be correctly enrolled in this system and won’t make duplicated certificate, better than just visit.

However, fingerprint system has its limits. It still could not solve the problem of working efficiency. For old people, they are not convenient in moving. Hence, a long-distance system should be made to save this kind of situation.

Luckily, the only remote biometric technology whose name is voiceprint recognition can meet this requirement quite well. It is very convenient. They don’t need to go to a fixed station to do survival authentication with this new system. Instead, make a call is enough to enroll to pension insurance system. Additionally, the processing time is short and the working efficiency will be higher.

There are companies that already made this proposal. I hope it can be put into use as soon as possible.

Source: (Chinese only)





Artificial Intelligence —- Build up from Small Pieces

Last blog, we talked about the intelligence of devices will increases in an unimaginative way. However, this need the effort from everybody, to build the real artificial intelligence one by one by small pieces of blocks.

Since an approach to artificial intelligence is simulating our brains, how do our brain thinking and processing. The idea will be that we program the small functions step by step as a block, without realizing what it will real be. For example, we program a small function that can only recognize circles, and it simulate the way how people thinking, like what’s the color of this circle, is it a regular circle or is it hidden of some parts?

This small program function should simulate whole path that human thinking as more as possible, if during the path there is somewhere too hard to achieve, we can divide it again, until we can easily without too much efforts to program it.

We don’t have to think will this program can be developed as a product or not, the only thing that we should do is to “Simulate” or brains!

“Divide and conquer.” As the old saying, we accumulatively build the blocks, and combine them, finally it would be an enormous system that can “Think itself”.

Phone Fraud and Speaker Recognition

Early in last month, a famous Taiwanese actress was in trouble with Phone fraud, she lose over 1 million dollars. This is not the only case I’ve heard. Just in last year, another famous actress from Hong Kong, been cheated over 0.1 million dollars. The scammers gave them telephone call and fraud their money. For us, this may seems so ridiculous, but it happens a lot, normal people may also encounter those problems but they just may just feel ashamed to share with others.

As far as I am concerned, most of those cases didn’t have a good ending, the scammers just escaped with Ill-gotten gains.

However, in February of 2015, a special news attract my eyes. It says that the police of HuaiYin cracked a Phone fraud, the suspect is a male who disguise as a female in the other end of phone. The technology of speaker recognition make big contribution to the success of arrested him.

As we can see from this successful case, speaker recognition is really becomes a very important tool of Identity verification. Fingerprint technology is already very mature and been used in so many aspect in our daily life. However, every technology will have its limitation. In dealing with telecommunications fraud, voiceprint technology plays a more important role.

Since speaker recognition technology is still not a very mature technology, there are many people who don’t trust it. But I think with the development of this technology and artificial intelligent, speaker recognition will for sure be widely used and facilitate our life.




Information Security Problem

After the information revolution, information becomes the key world of 21 century. Besides, information security will be also more and more important as a consequence.

Information security can be considered as keep information safe, cannot be get without authorization at beginning time where internet is just set up, only slow information will be transport on the lines, no wireless, no clouds and no social websites. At that time, it seems a very good encryption algorithm can solve information security problem.

Then when PC more and more common, virus play a role to steal information, they spread fast through internet. Then firewalls and virus scanner become popular.

However, nowadays, in my opinion, information security is not just keep the information impossible to be stolen, but make sure information don’t be abused without any authorization. Because clouds and social websites are increase so fast, they collect billions of billions user profiles, pictures, videos and so on. But some company just use those information for illegal usage.

How to solve those problems? First, keep the research on data encryption, develop good algorithms for data recognition (security recognition for speaker recognition as an example). Second, complete laws that should follow the changes of the changing world and suited for new up coming technologies. Third, we should be aware of those problems ourselves as a user, keep your information safe at your own side. Finally, I think the most important is that set up an organisation that for product safety appraisal, not only on the hardware side, but also the algorithms, the database for storing information, etc. This organisation should be supranational, should be organized by people from everywhere in the world. Every product come to the market should first go the appraisal procedure, and also appraisal annually. A appraised product will have a sign on it and users only choose the product with sign.

In conclusion, I would like to suggest the final proposal mentioned above. But all those methods are in the physical level, what we should do finally is improve the education level of human, upgrade the mentality of human being, let everyone do their effort to create a wonderful world without crime.

Comments and Questions from Previous Posts

In my previous blogs, they are a lot of comments about the security and difficulties of the speaker recognition. It’s time to gather all the information which I know to my followers. For example, a most common question is what if someone use a recording of the owner to unlock a door? Furthermore, if someone is very good at mimic others voice, can he unlock others voiceprint lock very easily?

A more difficult question is about using TTS (Text-to-speech). TTS is a speech synthesis system. If you have ever used google translate and click the speaker icon to listen to the voice of the text, you just used the TTS system. TTS converts normal language text into speech. The problem is that someone may record your voice to build your model and try to use a TTS speaker to unlock your voiceprint lock.

Last time, I was recommended to view a web page whose content is about security of your voice. The name is “Watch what you say, your smart-TV is listening”. It says the Samsung and Tesla smart-TV can spy on their users by voice recognition. I think this is really unsafe for the users.

I’m really curious on how the corresponding companies will solve those problems. Technically, how they build an anti-recording or anti-TTS speaker systems. In the customers side, how can they prevent their privacy not be misused by smart-TV companies or others.

If you can answer those questions in a technical way, or if you have any opinion about voiceprint privacy or related security problems, you are free to write your comment below and communicate with people interested in this topic.

Source: https://otlagenda.wordpress.com/2015/02/15/watch-what-you-say-your-smart-tv-is-listening/


How to Read a Paper in an Efficient Way

Our thesis is quite a theoretical thesis where full of reading papers. After reading some papers, I find that we don’t have to read word by word, and it won’t give us better understanding at all, sometimes it gives negative effects though.

How do we read the paper? I have some opinions:

1. Read the abstract and first 1 or 2 sentences of each paragraph in introduction part, to grasp the general idea of this paper. The reason why you should know the idea is that you want to know if this paper is really what you want, and can you find the answers here.

2. If this paper is suited, then read the equations and figures in background knowledge part. We you reading an equation, think about are the parameters a scalar or a matrix, and what is the dimension of the matrix, how can I get the parameters, and is that constant or an input or output. Flow chart is very important since it gives you better understanding to know the procedures.

3. Don’t afraid of the terms that you don’t know, skip it at beginning. Don’t let them become a barrier  to understand the main idea. Reading the references or search it on the internet if you really need it to understand some knowledge or equations.

4. Looking for some implementation of the theory, try to map the theory into practical things.

5. Read again and again if the paper is very useful!

That’s my opinions, I will add more if I come up with some new.

Thread of Artificial Intelligence

I read an article these days, which is about the thread of artificial intelligence. The idea of that article is basically about the growth curve of intelligence, which is a logarithm curve, and our human are in the time of the knee point where is the start of very fast intelligence increasing period.[1]

There is a simple reason that with the higher and higher processing speed of machines, we can do more and more calculations, and we aim to make them as smart as possible. However, we can already feel that, faster machines increase our working efficiency very much, it means that we can already do something that make the world more intelligent in an efficient way. We cannot imagine, how fast will the intelligence increases as if all the machine can also “Thinking”.(The real artificial intelligence).

But why fast intelligence increasing is not always good? We just afraid that the speed is too fast to be controlled by our human being. We afraid that robot will control our human just like in the science fiction movies, because we prefer sustainability, we prefer stability.

However, in my opinion, the real thing we should afraid of is the loss of affection between human and the degeneration among human. Because it is so common that people play their phones during a meal instead of chatting with people, that people getting use to shopping online without hanging a friend’s hand and people don’t even want to go to class and put a robot there, being a virtual themselves and being controlled by their master at home.[2](Luckily it’s only for sick student now)

All in all, let the intelligence grow fast, but don’t let our mind working slowly. Put away your phone when talk to a real people and balance the dependency between robot and your own body.

[1] “The AI Revolution: The Road to Superintelligence”, http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html, Tim Urban,

[2] “SICK STUDENT ATTENDS CLASS USING REMOTE-CONTROL ROBOT”, http://abc7.com/technology/sick-student-attends-class-using-remote-control-robot/402777, Eyewitness News, 2014.