On the fifth floor of the building of the Academy, Meng Fanqi and Han Ci had just come out of the dean's office.

"Have a meal, please." Han Ci turned her head to look at Meng Fanqi after pressing the elevator, she felt that since she wanted to trouble others, of course she couldn't let others do it in vain.

"You don't have to be so polite." Meng Fanqi knew what she meant, "If you hadn't mentioned my matter to Dean Fu at that time, Dean Fu wouldn't have noticed me in the library." ”

"His guidance is still very crucial to me, and some of the credit for this credit must be attributed to you." Meng Fanqi is sincere, if it weren't for Dean Fu's help, the quality of his early papers would definitely drop by one or two grades.

"I have merit, then you invite me to eat, I'm hungry." Han Ci never played duplicity and Tai Chi, since the other party said so, she just accepted it gladly.

This guy is really welcome, Meng Fanqi complained in his heart, but no matter what, it is at least very easy to get along with such a person. There won't be other ideas to hide, leaving you to guess all day long.

"At that time, I just thought your research in this area was somewhat interesting, and when I visited Senior Brother Fu, I mentioned things related to teachers and advocated paying attention to big data and artificial intelligence." Han Ci recalled the situation at that time, "I didn't expect that the content would be such a breakthrough achievement, according to my personal opinion, residuals are indeed not a very new thing in mathematics and physics." ”

"You're going to talk big, how much new is there under the sun?" Meng Fanqi knew that she didn't mean to belittle herself, and responded with a smile, "The content about the structure of the neural network itself is actually only a small part in the direction of AI. Residuals are just a trick to make the network easier to optimize, and from the mathematical point of view, it's certainly not a big deal. ”

Similar forms of residuals can be found everywhere in mathematics, such as Euler's forward and Newton's methods.

"At the end of the day, AI is a highly practical discipline. It is weak in theory and explainability, but its performance has improved by leaps and bounds, and it is of great value for industrial applications. I think that both the practical and the theoretical schools need to understand this matter objectively, and don't despise it all day long. ”

Meng Fanqi felt that Han Ci had already begun to have a bit of the taste of theory first. "Personally, I think it's more important to be able to actually impact people's lives, to feel your own results, and to leave the real foundation of theory to a genius like you to study it."

It's not that Meng Fanqi doesn't want to advance the progress of AI theory, but until 2023, there has been no progress in the principle and explainability of AI. Even Academician E, after a period of time explaining from the perspective of the dynamical system, and making some theoretical contributions, also diverted his attention away.

"Come on." Han Ci rolled his eyes, "I said don't engage in contempt, but it's not that our research has no impact on people's lives. ”

"It's not that your research will not have an impact on life, but in terms of this discipline and direction, it may be difficult to produce theoretical results. Even if it comes out, it is easy to match the experimental results. ”

Meng Fanqi is not light on theory and heavy on experiments, but since the take-off of AI, there are really too many unsolved mysteries that cannot be answered. The theories and laws of traditional machine learning have failed too much.

How many people have jumped into the big pit of AI explainability one after another, but in the end, most of them are far-fetched explanations, which is a bit of a sense of pointing somewhere.

On the contrary, some practical work that focuses on visualization is very helpful in understanding the patterns of AI models.

"It's more driven by curiosity, I didn't expect a world-class breakthrough in the technology you wrote in the library. Now that I know its value, I am really curious. Han Ci was actually the second person in the world to know about the residual network, but she only read it cursory, glanced at the summary, and didn't have a chance to read it carefully, and didn't pay much attention to it at all.

This is also normal, who would have thought that they would meet someone in the library who has broken the world algorithm performance record by a large margin?

"Speaking of which, your DreamNet paper hasn't been published yet, has it? When will the plan be sent? Dean Fu didn't reveal Meng Fanqi's original manuscript to others, but he still talked a lot about the specific content with his little junior sister.

"Not yet, I'm going to announce it when I go to Australia next month." Meng Fanqi was not in a hurry about this matter at all, he guessed that Dean Fu would not show his manuscript to others, so he specially asked himself to answer some questions from Han Ci.

He doesn't really care personally, the importance of articles, codes and experimental results in the direction of artificial intelligence is almost the same as the theory and the article itself.

Even Meng Fanqi, who had been exposed to the source code and a large number of experiments, took more than half a year to reproduce these contents, and there was no risk of theft.

So Meng Fanqi took the original manuscript from his bag to Han Ci quite happily.

"Don't you wonder why I care so much about your work?" Han Ci took the manuscript of the paper and quickly skimmed it as he walked.

"I probably want to start with the relationship between DreamNet's residuals and mathematics." Meng Fanqi did not pay much attention to theoretical work, but he still knew a little about some of the ideas of the academician school.

I know what it is, but I don't know why. But at this point in time, it must be enough to fool Han Ci.

"Oh? Tell me more about it? Han Ci was really a little surprised at this time, Dean Fu liked this little junior sister very much, so he had a lot of exchanges with her. There are also one or two mentions about Meng Fanqi.

In the exchange between the two, Han Ci's feeling is that Meng Fanqi's judgment and vision of the direction of artificial intelligence are very good, and he has a wide range of interests and is biased towards practical technology.

The code ability is very good, and the theories and algorithms proposed by myself can be implemented with relatively high performance, but the definition and deduction of theory and mathematics can only be said to be a very ordinary level.

In the eyes of the protégés of the two famous academicians of mathematics, nature is also the posture of the midstream, and there is nothing particularly commendable.

Han Ci fully admits and admires Meng Fanqi's talent in artificial intelligence, but when it comes to mathematics and physics, he really didn't expect him to have any unique insights.

At first glance, Han Ci didn't believe it, so he asked him to continue talking.

"There should be a lot of directions that can be done in the residual network from the perspective of the dynamical system." Meng Fanqi smiled secretly, do you still want to test me? I know what papers your teacher will send in the next two years.