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
Is AI deep learning entry barrier low?
Is AI deep learning entry barrier low?
For a long time, I have noticed that many people who claim to be deep learning experts and big coffees are actually not worthy of their names. These people don't have a machine/deep learning education or research background, just install TensorFlow and run some code on GitHub, then they will recognize experts, write blogs, write tutorials, and even publish books.
This made me very disturbed, which destroyed the reputation of deep learning. Most companies don't know how to distinguish these so-called "experts". Interviewers don't understand deep learning, and don't care about NIPS and ICML. So when these "experts" deep learning programs don't work, these companies will think that everything is just a hype.
With this situation increasing, more and more people are suspicious, and even the insiders have begun to discuss bubbles. How do you see this question? Do you agree with my opinion? What should I do in the future?
640?wx_fmt=jpeg
Many people follow the arguments against the above arguments. The central idea can be summed up as: research more questions and talk less about politics.
Contrary to the concept of the po master, openness or “low threshold” is one of the best parts of the machine learning community. This community will not shut people out of their identity and will be happy with everyone.
Most companies need engineers, not researchers. Most of today's work is based on specification techniques and then applied to specific businesses. No need to do any new research.
According to my experience, a good professor can provide a lot of good ideas and make more than 20 excellent engineers busy. This match is appropriate. Usually at least one (and possibly only one) excellent professor and a team of smart engineers are needed to support him.
Who knows how to apply machine learning to handle business needs? Who defines the input and output of the model? Who guarantees the availability of data? Who analyzed it but what happened? Who handled the problem?
If the engineer has done all this, it is not a simple engineer. You will always need those who can insight into business needs, map them to the right questions and answers, and turn them into people who rely on consumer code.
Oppose this classification and opposition. Many researchers are also very good at applying these knowledge in the ideal world. However, there are ways to use machine learning to deal with problems, and to say that my machine learning experts are completely different.
This debate is largely due to the lack of necessary recognition by engineers, which has led some people to "disguise" into scientists. Not submitting papers to NIPS or ICML does not mean that they do not understand the underlying principles of deep learning. Admittedly, even if the entry threshold is lower, there are still not enough deep learning engineers to meet the demand.
This category is in urgent need of experienced machine learning engineers.
This made me very disturbed, which destroyed the reputation of deep learning. Most companies don't know how to distinguish these so-called "experts". Interviewers don't understand deep learning, and don't care about NIPS and ICML. So when these "experts" deep learning programs don't work, these companies will think that everything is just a hype.
With this situation increasing, more and more people are suspicious, and even the insiders have begun to discuss bubbles. How do you see this question? Do you agree with my opinion? What should I do in the future?
640?wx_fmt=jpeg
Many people follow the arguments against the above arguments. The central idea can be summed up as: research more questions and talk less about politics.
Contrary to the concept of the po master, openness or “low threshold” is one of the best parts of the machine learning community. This community will not shut people out of their identity and will be happy with everyone.
Most companies need engineers, not researchers. Most of today's work is based on specification techniques and then applied to specific businesses. No need to do any new research.
According to my experience, a good professor can provide a lot of good ideas and make more than 20 excellent engineers busy. This match is appropriate. Usually at least one (and possibly only one) excellent professor and a team of smart engineers are needed to support him.
Who knows how to apply machine learning to handle business needs? Who defines the input and output of the model? Who guarantees the availability of data? Who analyzed it but what happened? Who handled the problem?
If the engineer has done all this, it is not a simple engineer. You will always need those who can insight into business needs, map them to the right questions and answers, and turn them into people who rely on consumer code.
Oppose this classification and opposition. Many researchers are also very good at applying these knowledge in the ideal world. However, there are ways to use machine learning to deal with problems, and to say that my machine learning experts are completely different.
This debate is largely due to the lack of necessary recognition by engineers, which has led some people to "disguise" into scientists. Not submitting papers to NIPS or ICML does not mean that they do not understand the underlying principles of deep learning. Admittedly, even if the entry threshold is lower, there are still not enough deep learning engineers to meet the demand.
This category is in urgent need of experienced machine learning engineers.
PREVIOUS:Business license OCR
NEXT:H5 OCR identification identification