Speaker Recognition, Still a Long Way to Go

For the last post I will make for our blog, I want to say a little bit about the recent condition and the future about speaker recognition.

Although speaker recognition have already been researched for several decades, it still have a long way to go. We have good algorithms that can get the recognition accuracy to more than 98%, even 100% for very long time training. Those prototypes are more aiming to high performance instead of computation cost and so on. So computation time is a barrier that stop the system to be used in some real system like banking system etc. An 630 people’s group needs few hours to generate the model, we can’t imagine how many time is needed for a bank where have millions of people. But at least you can divide the calculation in different processors which however needs investigate more money.

Another barrier is that the high performances are getting from very clean speech where don’t even have silence or other voice instead of the speaker. However, in real application and devices, we are difficult to filter out all the noise and other sounds, these will give a big challenge to the frond-end processing and the innovation of good recording hardware.

The biggest problem will still be the security problems, since we can simply mimic a people’s voice or record that, that will make the speaker recognition system more complex to detect the channel difference from the source and receiver. Some sophistic algorithms will be involved in.

So, a very good speaker recognition system needs enormous peripherals, and the balance between the core system and the outside processing parts. All in all, it seems we still need a long way to go to achieve a simple, accurate, secure speaker recognition system!

Be aware of computer security

Again, we back to security problem. Computer virus seems not a big issue for us, if we correctly use firewalls, secure protocols and secure transmission methods, we can keep virus free roughly. However, computer security is more broader since everything electronics nowadays use at a processor and it can communicate with outside somehow.

I saw a video about computer security[1]. In that video, there are several experiments they did to revel the possibilities that it is easily to hack an electronic device. They hacked a healthy device which is putted in human’s heart by wireless, and control the action of it. Nowadays, almost all devices in body are using wireless for communication, what will happen if we use wireless to wantonly control those devices?

Same concept was using on cars also, they use wireless signal and find bugs in software in cars, and use those bugs, they can plant their software in. They can control the break, the FM, the dashboard, etc. And they can also steal a car by this.

When we start being aware of camera security, they have already found that they can also use the accelerator in phones to detect what people type when their phones are beside their keyboards. After learning some human common typing habits, they success using iphone 4 to find out what people type.

Those experiments are astonishing me, they really happen in our life, as long as an experienced hacker, he can use those method hack our devices. It make me thinking, besides we need be aware protecting our information security by know the truth and action against it, we also need let everybody be aware of educating people don’t do criminal things.

[1] “All your devices can be hacked”, Avi Rubin, TED video, https://www.ted.com/talks/avi_rubin_all_your_devices_can_be_hacked

Can you beat AI?

Google recently built an AI that can learn how to play a Atari video game itself, after a while of training, it can master the game. The most impressive part might be that the AI not only matched wits with human players in most cases, but actually went above and beyond the best scores of expert meat-based players in 29 of the 49 games it learned, and bested existing computer based players in a whopping 43.[1]

This experiment really aware me, although AI is still in the beginning part, it can already beat human being. AI play a game just a prototype, it will be used in military, in transportation, in medical, in every parts in our life. If it goes good direction, maybe it can give more accuracy surgery than a human doctor, or it becomes a trusted friend and entertain us during tedious works. However, if AI weapons is created for conquer and war, then it will destroy human, and finally it will becomes the war of AIs, maybe the scenario in science fiction movies will come true.

But we can’t stop the development of AI, we need the growing technology. Besides, at meantime we focus on the innovation of technology, we also need take a glare at ethics, before publish the technology, be aware of the influences.

[1] “Google Builds An AI That Can Learn And Master Video Games”, http://techcrunch.com/2015/02/25/google-builds-an-ai-that-can-learn-and-master-video-games/, Feb 25, 2015,Darrell Etherington.

Building “self-aware” robots

This is another video about Self-awareness robots. Those robots can self learn after its birth although they don’t know what is it looks like and what can it do at beginning. They just try and error, and learn step by step, finally they can find a way that can move forward or copy its self as long as you give it enough basic building blocks and energy.

The idea is same as previous post, that build self learn robots. In my opinion these kinds of algorithms is so called “Classification algorithms”, it collect different data from all kinds of sensors, and classify them into groups. Those groups can be seen as a new knowledge that the robot learnt.

The self learning idea give us a wider chance to achieve the real artificial intelligent. Along with more and more faster computation speed plus more and more varieties of sensors, robots might learn more themselves. That means intelligent will increase more and more faster (like a previous post said).

It seems that people still doubting will the real artificial intelligent will come true. However, this real product can be a very good proof that can confirm the answer that intelligent will increase exponentially and the real artificial will come true!

[1] TED vedio, “Building “self-aware” robots”, https://www.ted.com/talks/hod_lipson_builds_self_aware_robots

Robots with “soul”

I watched a video recently, it’s title is “Guy Hoffman: Robots with ‘soul'”. I really like the word “soul”, since I think only human being can have a “soul”, but how come a robot can with a “soul” in it?
“Soul” is a human made feature, it describe something with life. It has its own thought and idea, it won’t always act as you expected. A robot with “soul” is just those kinds of robots who can do what you can’t expect and interact with you.
In this video, the lecturer described several projects he made. In those projects, some robots can play an instrument with its own score, it can learn from ambiances, then creates its music which will perfectly match the whole concert. After the success, the lecturer added a “Head” and “Body” to his robot, and using some machine learning algorithms that let the robot can interact with the accompanists, like looking at each other, and nod its head following the melody. These “body languages” indeed influence the people around by proof of experiment. People feel excited working with those kinds of robots that with “soul”, and people call it “He or She” instead of “it”.

I think it is a good point of view that how to develop an algorithm for robots, give a “soul” to robot along with the beneficial functions, that will make the robot more real and more approaching to the real artificial intelligent.

[1] TED video, “Guy Hoffman: Robots with ‘soul'”, http://www.ted.com/talks/guy_hoffman_robots_with_soul#t-1034841

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”.