Compared with the previous English letters, this "Preface to the Lanting Collection" is more difficult, and has more stringent requirements for the brain-computer sensing system's ability to capture and identify brain electrical signals.
The first is the transition from single English letters to Chinese characters, especially the complexity of our Chinese characters, which is a very severe test for the accuracy of capture and recognition of the brain-computer sensing system.
First of all, the structure of Chinese characters is very complex. Not only are there many gestures, but they are also very difficult to write. Moreover, the diversity of Chinese characters with homophones, similar forms, synonyms, rare characters, and traditional Chinese characters also makes the test more difficult.
The words and sentences composed of these Chinese characters are even more complex.
How to make the brain-computer sensing system accurately recognize the words imagined in your brain requires relevant technical support.
One of the simplest methods is to use contextual association recognition technology similar to intelligent speech systems or intelligent input methods to determine the exact vocabulary and text in the sentence through association of a whole sentence.
This technology has largely solved the problems of vocabulary and Chinese characters with multiple sounds, multiple meanings, and similar forms, and improved the efficiency of input recognition.
If you just want to use an input method, then there is no problem in using this contextual association recognition technology in the field of brain-computer sensing technology.
However, the brain's ideas and thoughts are changing rapidly, and the use of this association recognition technology cannot keep up with the brain's thinking speed, and errors may occur.
Therefore, in the development of their project, Wu Hao asked the scientific research team to try to avoid using this contextual association recognition technology. It can only exist as a final optimization tool, not as the technical main body of this system.
When Cai Tao reads the original text of this article, each word is transmitted to the brain through the eyes and forms a specific brain electrical signal. This is the memory signal of this word in our brain and is captured by us.
recorded by the brain.
The brain-computer sensing system captures and recognizes this EEG signal and restores it to the text itself. This is the main content of this test.
To put it bluntly, in this test, the brain only serves as a relay transmission function. Don’t underestimate this function. How to ensure the accuracy and speed of these relay information is also a topic that the scientific research team needs to study, and it is also the main difficulty.
The second difficulty is interference, constant interference, especially when reading familiar articles, it is easy to be interfered by other incidental memories.
Just like when Cai Tao read this familiar preface to the Lanting anthology, he couldn't help but think back to his middle school life and some memory information closely related to this text. How to eliminate the interference of this memory information and accurately capture and identify this text?
The content of this text is also a very difficult problem.
The third problem is that when Cai Tao is reading this kind of familiar article, he will inadvertently speed up. This is true for reading, and the same is true for brain thinking. How to capture these EEG signals under such fast thinking is also a question.
A technical problem faced by the R&D team.
Of course, there are many more difficulties, so I won’t list them all here.
Sure enough, the results of this test were far less good than the previous test. In the "Preface to the Lanting Collection" produced by Cai Tao's EEG signals captured and identified by brain-computer sensing equipment, many errors appeared.
There are typos, wrong words, wrong sentences, and even some other content mixed in, etc.
Faced with such data, everyone can't help but feel a little frustrated. Sure enough, it still doesn't work.
Wu Hao certainly felt this atmosphere. He then smiled and clapped his hands and encouraged everyone: "Don't be discouraged. In my opinion, the test results have been very successful. Without the assistance of intelligent association recognition technology,
Next, we have been able to capture and identify an article, and the accuracy rate reached 87%, which is already very impressive.
It can be said that my test results are definitely far ahead in the world.
So, don’t be too discouraged, let’s keep up the good work. I believe that with the efforts of all of us, this technology will definitely flourish in our hands.”
"Mr. Wu is right. We should not be too sluggish. This technology has grown from scratch and has reached the point where it is now. I am proud of our entire team and each of us. Let's continue to work hard.
I believe these problems will be solved soon." Ding Zifei also opened his mouth to encourage everyone.
Sure enough, he had started his own business, and just a few words could cheer everyone up. Of course, what Wu Hao said before also played a crucial role. Otherwise, these people's emotions would not have been able to change in such a short period of time.
Good comfort.
"Okay, let's proceed to the next test content." Ding Zifei gave instructions and then introduced it to Wu Hao.
"This test content is the normal text input method. The intelligent Lenovo input method is used for input testing. This is also a content we often test. The accuracy can reach more than 99%, and the input speed is ours.
The keyboard input speed is two to five times faster, which can be said to be good news for word workers and the nemesis of tentacle monsters."
Wu Hao knew this test well. At previous artificial intelligence conferences, relevant research teams had conducted similar test sessions, using brain consciousness to perform typing.
The test at that time was very eye-catching and attracted the attention of the entire society. Some experts even said that brain-computer sensing equipment will soon be put into commercial use to provide medical equipment for patients with high paraplegia and ALS who are plagued by diseases, and even for vegetative patients.
Come to the gospel.
But after so many years, this technology has not been put into commercial use as the expert said.
Looking back from today's perspective, the conscious typing technology displayed at that time was still very rudimentary and very slow. It was basically spelled out through pinyin, letter by letter, and the efficiency was very low.
"The test begins, Cai Tao, write down the "Preface to the Lanting Collection" that you read in the last test content."
"OK!"
Then in the blank Word document, Cai Tao quickly started typing. The speed was very fast. The entire preface to the Lanting Collection, a total of 400 characters, was written down in less than thirty seconds. And the speed was very fast, just
I see that the input box is constantly flashing, and the text and vocabulary input cannot be clearly seen at all.
"27 seconds, equivalent to 14.814 words per second, which is 0.6 words higher than the last test. The result is good!" Ding Zifei looked at the data on the screen and said with a smile,
Bang bang bang bang...
Wu Hao took the lead in applauding, and soon the entire laboratory was filled with applause.
"Yes, I think this technology can be brought to the market for commercial use!" Wu Hao praised Ding Zifei while applauding.