Sha Mianzhi knew that Wang Hao was also going to make a report, and said that he had become a competitor with himself, and he suddenly felt a little nervous.

He used to be Wang Hao's teacher, but Wang Hao is no longer an 'impressed' undergraduate, anyone who can go to graduate school and doctoral degree at Donggang University can be called an elite talent, and a doctoral student who can stay at Donggang University can be said to be an 'elite among the elites'.

Even Xu Jie, who was the most optimistic doctor under his staff, brought over to participate in the STACS meeting, did not dare to say that he could stay at Donggang University 100%.

Donggang University, there are too many talents!

Wang Hao still belongs to a very outstanding kind, not to talk about the problem of materials laboratory, he left Donggang within a few months, he completed a number of SCI papers, there are several algorithm papers, all published in core journals, among which the paper of 'Fourier transform to build mathematical models' has had a great impact on the Internet industry, which can promote big data analysis to become faster and more accurate.

This level can be said to be top-notch.

However, it is precisely because of the public opinion some time ago that Sha Mianzhi knew that Wang Hao had just completed several studies, and when he first met Wang Hao, he didn't think that he was here to give a report, but he just felt like other scholars, who came to the STACS conference just to increase his knowledge and learn, but Ruan Hailong, who was with him, was more like he came to make a report.

After separating from Wang Hao and Ruan Hailong, Sha Mian and his party went to the conference hall again, and went to the electronic column to carefully look for Wang Hao's report.

"At the end, the second hall, the first session in the afternoon." Wang Mingkun took a quick look and found Wang Hao's report.

"Multiplication of large numbers?"

Wang Mingkun thought about it and said, "He has just completed the research related to the Fourier algorithm, and it is estimated that it is an improvement algorithm." ”

He estimated, "Small results, good luck, the manuscript is approved!" ”

"That's how it should be." Sha Mianzhi thought about it and nodded, he had just completed a major study, normally, it is unlikely to complete another one quickly, the large number multiplication algorithm is a focus problem, but there has been no breakthrough for decades, and for the improvement of the algorithm and the improvement of efficiency in specific situations, it can only be classified as a 'small research'.

"Wang Hao, it's probably just a joke!" Sha Mianzhi suddenly smiled and shook his head.

Xu Jie next to him suddenly asked, "Teacher Sha, I don't really understand, isn't Wang Hao a doctor of mathematics?" How did he come to a computer conference? ”

Before Sha Mianzhi could speak, Wang Mingkun smiled and said, "Xiao Xu, if you want to go further in the basic research direction of computers, you must learn mathematics well, and it is best to be able to learn it profoundly." ”

"Mathematics is the foundation of the computer, no matter what the algorithm is, the foundation is mathematics."

"If you are good at mathematics, it is easy to achieve results in computer research, on the other hand, you are just good at computers, you know a lot, and you can use it, you can only be a programmer, and there is no way to engage in the research of theories and algorithms."

Sha Mianzhi also smiled, "Listen to what Teacher Wang said, that's why I let you watch more math." ”

Xu Jie nodded knowingly.

When I think of Wang Hao, I can't help but feel a little envious, the other party is one year younger than himself, but he has already completed influential research, and can give a report at a top conference, and can have an equal dialogue with Sha Mianzhi and Wang Mingkun.

And he is just a doctoral student who follows his supervisor.

……

The next day, the meeting officially began.

Wang Hao and Ruan Hailong came very early, there were no special seats arranged at the venue, and they didn't do too far ahead, but just found a place in the middle.

Not long after, the three of Sha Mianzhi also arrived, so they walked over and sat down side by side.

The morning presentations were all conducted in Room 1, and it was the turn of some 'unfavorable' reports to be conducted together in both rooms.

After the opening time, the organizers of the meeting, the chairman, and several committee members entered the meeting one after another, and then the chairman came to the stage and said a lot of words, which can be simply summarized as 'the meeting officially began'.

The opening of the STACS meeting was not that complicated, but the Chair said a few words, summarized last year's meeting, and then said about the schedule of this year's meeting, and then announced that the report meeting was officially opened.

The first to take the stage was a young professor, probably only about 30 years old, from École Polytechnique.

The meeting held by Qi Guo must take care of 'its own people', and it also uses a 'not very important but not faulty' study as the opening remarks to officially put the meeting on the right track.

The young professor talked about the research on the 'online algorithm-regularized double average algorithm', which sounds a bit interesting, but the research has only made some progress, and it is difficult to attract the attention of the scholars present.

Twenty minutes, the report ends.

Some people in the room applauded politely, but most of the applause came from the front row, from the organizers, judges, and invited experts, and the scholars in the back row did not even have a symbolic applause.

They are already very polite to the research that is 'not interested' and 'meaningless', and does not shout 'go down' or 'go down'.

This is the real academic conference.

If you have the strength, you can win applause, and if you don't have the strength, don't go up at all, scholars don't understand 'hypocritical politeness'.

Wang Hao also listened to the first report a few times, and found that it was not helpful to him, and the research and development progress was also very small, there was nothing more outstanding, just like others, there was no interest.

When it came to the second report, he listened very attentively and patiently, and even used a 'teaching coin'.

Others listened attentively, too.

The first report is equivalent to 'making an opening', and the second and third reports are more important and are studies that the reviewers of the venue consider to be of 'great significance'.

Many scholars who do not need to make presentations come to the conference for the purpose of listening to research of great significance, professional top conferences, and also a place to increase knowledge and learn things, new research directions, new content, so that scholars can know what others are researching, how to complete the research, so that they can open up their thinking and find inspiration and direction related to their own research and development.

Wang Hao attaches the most importance to the three reports, today's second and third sessions, and tomorrow morning's third session, that is, Sha Mianzhi and Wang Mingkun's research together.

The second one is a study by a professor at the University of Oxford, which is a theoretical study of the computational complexity of gradient descent algorithms.

This is very rare.

Much of the applied research relies on an algorithm called 'gradient descent', which is a process of solving the maximum/minimum value of a mathematical function, from calculating the optimal way to produce a product to the best way to schedule workers' shifts.

However, compared with multi-directional applications, related theoretical research is scarce.

The professor at the University of Oxford, who gave the report, studied the intersection problems in various situations in the form of mathematical computers from the two directions of "gradient descent algorithm does not work well on many common problems" and "a lot of work on gradient descent does not involve complexity theory", so as to theoretically demonstrate the gradient descent algorithm.

Wang Hao listened with relish, and the clear logical analysis in the argument made him feel that his grasp of the logical argument was clearer.

The other reaction is --

[Task 2, inspiration +1.] 】

After listening to the report of the whole process, it directly brought a little inspiration to 'Task 2', and it was obvious that the harvest was huge.

Although there is only a little inspiration, you must know that 'Task 2' is to crack the number of gods, and the difficulty is A-level, just adding a little inspiration may be a very important improvement.

The professor's report at the University of Oxford was unanimously praised and received unanimous applause when it was completed.

Sha Mianzhi sat next to Wang Hao, and couldn't help but sigh, "It seems that it's not easy to get the best!" He is confident in his research, but it is difficult to say that it is difficult to suppress the report just now, and it still depends on the opinion of the conference review team.

Here's the third one.

The second and third sessions of the first day of the conference can be said to be the finale, and with the wonderful report just now, many people are also looking forward to the third one, which will be played by Simis Gollix from the University of Helsinki in Finland, and his colleague Almalov.

The name of the report is 'Fast and Accurate Minimum Mean Square Solution', and it is an improvement of the minimum mean square algorithm, also known as the LMS algorithm.

When Simis Gollix came to power, he proudly announced, "We have found the fastest and most accurate method for solving the minimum mean square, which can reduce computational complexity by more than two orders of magnitude without sacrificing accuracy and improved numerical stability." ”

This sentence immediately caused an uproar in the venue.

Solving the minimum mean square is at the heart of many machine learning algorithms, and it is no joke that it can reduce computational complexity by more than two orders of magnitude.

For example, it takes 100 million operations to calculate a problem, and a drop of two orders of magnitude becomes a million.

This is clearly a qualitative leap.

Simis Gollix began to explain in earnest, and his colleague Almalov supplemented by a very novel division and conquer method, and then used the discrete Fourier transform algorithm to act as the 'helm' of the overall architecture.

Wang Hao suddenly felt energetic when he heard this.

He felt that the content of 'helming' seemed familiar, and he understood it if he continued to listen to it.

The following research content must have some reference for his own 'Fourier transform-assisted construction of mathematical models'.

"Plagiarism?"

"No, it should be said that it is an application."

It is also normal for the content of the published paper to be used as a reference for other research, as long as the paper has a 'reference'.

And that's okay.

However, Wang Hao continued to listen, and he couldn't help frowning, he found that the other party's research was problematic, especially when it involved the discrete Fourier transform algorithm, which combined the 'divide and conquer method' to support and reduce the computational complexity, and there was a problem when it reached two orders of magnitude.

After about an hour of presentation, Simis Gollix finished most of the explanation, he explained the 'general direction', and also stopped to rest, and let everyone in the room digest it.

Everyone else was amazed at the results of the report, but Wang Hao shouted, "Mr. Golliks! ”

Simis Golix immediately noticed Wang Hao and asked suspiciously, "What is the problem with this young gentleman?" ”

Everyone in the venue immediately looked over.

Wang Hao stood up and said, "Your report is wonderful, I mean the front, but the second part, using the discrete Fourier transform to frame the overall calculation, I think, is problematic. ”

"The combination of the discrete Fourier transform and your 'divide and conquer' method makes it impossible to include all the solutions when calculating super-large numbers or super-large quantities, say, more than trillions of calculations, shaping a kernel set of complexity."

"You rely on the Caratheodory theorem to complete the construction, and there is no problem with the divide and conquer method itself, but when combined with the discrete Fourier transform algorithm, there will be problems."

"It is in the second part that the characterization of the convex point begins ......"

Wang Hao pointed in a direction.

Simis Golix didn't look back, but said with disdain on his face, "Young man, what about your mentor?" ”

"This is a STACS meeting, and you're responsible for what you say."

Wang Hao chuckled, "Of course I will be responsible for my words." Also, I came to give a presentation, and I wasn't with the 'mentor'. The last sentence was made of teasing.

He continued, "And, if I heard me right, from the second part onwards, your research has mainly used my methods. ”