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
99% of the companies are in contact with artificial intelligence AI, who can win the Google?
99% of the companies are in contact with artificial intelligence AI, who can win the Google?
Even with the support of artificial intelligence frameworks such as TensorFlow or OpenAI, artificial intelligence still needs deep knowledge and understanding compared to mainstream web developers. If you ever set up a work prototype, you might be the smartest person in the room. Congratulations, you have become a member of the senior club.
On Kaggle, you can even make considerable profits through projects that deal with the ideal world. In a word, it's a good job, but can it be enough for you to build an enterprise? After all, you can't change the market mechanism. From a commercial point of view, artificial intelligence is just another way of implementing the existing problems. Customers don't care how to do it. They only care for the results. This means that you can not only through artificial intelligence to boon. When the honeymoon is over, you need to invent the value. From a long distance, as long as the customer is the most important.
Although your customers may not be interested in artificial intelligence, Vic is very concerned about it. The news media are also concerned. This is the case in many industries. This difference in attention may give the venture company an ideal distortion to create a risk, but don't be fooled: unless you have invented the general AI, there will be no free lunch. Even if you are the favorite of the venture, you must go to the end of your customer. So let's also be a driver and see how we can prepare for the future.
"Mainstream artificial intelligent train"
Artificial intelligence seems to be different from other big trends, such as block chain, Internet of things, financial technology, and so on. Of course, its future is unpredictable. But it's all the technology. The difference is that we seem to be at risk as a person's value proposition - not just other industries. The value of our decision-makers and creatives is being reevaluated. This has aroused the emotional repercussions of people. We don't know how to locate myself.
A very limited number of foundation technology, most of them can be classified as "deep learning", which constitutes the basis of almost all applications: convolutional neural network and short memory network, automatic encoder, random forest, gradient strengthening technology, and a few other applications.
Artificial intelligence also provides many other ways, but these central mechanisms have recently won an overwhelming victory. Most researchers think that the future of artificial intelligence technology is coming from the improvement of these technologies (rather than those that have substantial differences with them). For these reasons, we can call this "the mainstream artificial intelligence discussion".
Any real world processing plan is composed of these center algorithms and non AI shapes, to prepare and process data (for example, data preparation, functional engineering, environmental modeling). The local improvement of artificial intelligence often makes the part of the non artificial intelligence redundant. This is the essence of artificial intelligence, and it is simply the definition of it - the way to deal with specific problems is out of date. However, the part of this non artificial intelligence is usually the real profit source of a company driven by artificial intelligence. This is their secret weapon.
Every improvement in artificial intelligence makes this competitive advantage more likely to be open source, and everyone can use it. But it also brings disastrous results. As Frederic Jelinek once said, "every time I quit a speech scientist, the performance of the speech recognizer will be improved."
Machine learning has fundamentally introduced the next stage of layoffs: the code is simplified to data. Virtually all of the identification techniques based on models, probabilities and rules are eliminated by deep learning algorithms in 2010.
Now only a few hundred lines of scripts (plus a considerable amount of data) can beat category expertise, function modeling, and thousands of lines of code. As mentioned above, this means that proprietary code is no longer a defensible asset on the track of a mainstream artificial intelligent train.
Serious dedication is extremely rare. A real break or a new pause, even a new part of a local combination, can be achieved only by a very limited number of researchers. As you might think, the inner circle is much smaller (the number of developers must be less than 100).
Why is this? Perhaps this is rooted in its central algorithm: backpropagation. Almost every neural network is exercised in this way. The simplest way to reverse the reverse communication can be learned in the first semester's calculus course, and the integrity is not complex (but not in primary school). Though it seems very simple -- or maybe for this reason -- in the colorful history of over 50 years, only a few people can see the difficulty and question its main structure.
If backpropagation can be visible as it is today, our achievements may be 10 years ahead of the present phase (except for Computing).
From the general neural network of 70s to the recirculation network, to the long and short term memory network, the category of artificial intelligence has been vibrated. And it only needs dozens of lines of code! Several generations of students and researchers calculated the gradient descent method through mathematical calculation, which proved its correctness. But at the end of the day, most people nodded, saying, "this is just an optimization way," and continue to work hard. It is not enough to dissect understanding. You need some kind of "creative instinct" to make it different.
It's not easy to have the top level of research in the industry, so 99.9% of the companies are just the mainstream people.
On Kaggle, you can even make considerable profits through projects that deal with the ideal world. In a word, it's a good job, but can it be enough for you to build an enterprise? After all, you can't change the market mechanism. From a commercial point of view, artificial intelligence is just another way of implementing the existing problems. Customers don't care how to do it. They only care for the results. This means that you can not only through artificial intelligence to boon. When the honeymoon is over, you need to invent the value. From a long distance, as long as the customer is the most important.
Although your customers may not be interested in artificial intelligence, Vic is very concerned about it. The news media are also concerned. This is the case in many industries. This difference in attention may give the venture company an ideal distortion to create a risk, but don't be fooled: unless you have invented the general AI, there will be no free lunch. Even if you are the favorite of the venture, you must go to the end of your customer. So let's also be a driver and see how we can prepare for the future.
"Mainstream artificial intelligent train"
Artificial intelligence seems to be different from other big trends, such as block chain, Internet of things, financial technology, and so on. Of course, its future is unpredictable. But it's all the technology. The difference is that we seem to be at risk as a person's value proposition - not just other industries. The value of our decision-makers and creatives is being reevaluated. This has aroused the emotional repercussions of people. We don't know how to locate myself.
A very limited number of foundation technology, most of them can be classified as "deep learning", which constitutes the basis of almost all applications: convolutional neural network and short memory network, automatic encoder, random forest, gradient strengthening technology, and a few other applications.
Artificial intelligence also provides many other ways, but these central mechanisms have recently won an overwhelming victory. Most researchers think that the future of artificial intelligence technology is coming from the improvement of these technologies (rather than those that have substantial differences with them). For these reasons, we can call this "the mainstream artificial intelligence discussion".
Any real world processing plan is composed of these center algorithms and non AI shapes, to prepare and process data (for example, data preparation, functional engineering, environmental modeling). The local improvement of artificial intelligence often makes the part of the non artificial intelligence redundant. This is the essence of artificial intelligence, and it is simply the definition of it - the way to deal with specific problems is out of date. However, the part of this non artificial intelligence is usually the real profit source of a company driven by artificial intelligence. This is their secret weapon.
Every improvement in artificial intelligence makes this competitive advantage more likely to be open source, and everyone can use it. But it also brings disastrous results. As Frederic Jelinek once said, "every time I quit a speech scientist, the performance of the speech recognizer will be improved."
Machine learning has fundamentally introduced the next stage of layoffs: the code is simplified to data. Virtually all of the identification techniques based on models, probabilities and rules are eliminated by deep learning algorithms in 2010.
Now only a few hundred lines of scripts (plus a considerable amount of data) can beat category expertise, function modeling, and thousands of lines of code. As mentioned above, this means that proprietary code is no longer a defensible asset on the track of a mainstream artificial intelligent train.
Serious dedication is extremely rare. A real break or a new pause, even a new part of a local combination, can be achieved only by a very limited number of researchers. As you might think, the inner circle is much smaller (the number of developers must be less than 100).
Why is this? Perhaps this is rooted in its central algorithm: backpropagation. Almost every neural network is exercised in this way. The simplest way to reverse the reverse communication can be learned in the first semester's calculus course, and the integrity is not complex (but not in primary school). Though it seems very simple -- or maybe for this reason -- in the colorful history of over 50 years, only a few people can see the difficulty and question its main structure.
If backpropagation can be visible as it is today, our achievements may be 10 years ahead of the present phase (except for Computing).
From the general neural network of 70s to the recirculation network, to the long and short term memory network, the category of artificial intelligence has been vibrated. And it only needs dozens of lines of code! Several generations of students and researchers calculated the gradient descent method through mathematical calculation, which proved its correctness. But at the end of the day, most people nodded, saying, "this is just an optimization way," and continue to work hard. It is not enough to dissect understanding. You need some kind of "creative instinct" to make it different.
It's not easy to have the top level of research in the industry, so 99.9% of the companies are just the mainstream people.