179: Chapter 179 Public Opinion Explodes Again!

"It seems that once this paper is published, it will definitely set off a massive wave throughout the entire computer science community!"

After thinking for a moment, Charles couldn't restrain his urge to express himself. He opened his personal Twitter account and his frequently used academic blog homepage.

As a senior editor for NIPS and a well-known scholar in the industry, Charles's social media had great influence and a large following of professional fans in the field of computer deep learning.

Reading selected Taiwan novels in his spare time was quite pleasant.

He quickly typed out a passage in English and clicked send.

"Just reviewed an incredible submission in the NIPS backend. The gradient staleness problem in Asynchronous SGD (Stochastic Gradient Descent) may have been completely solved by a brand-new mathematical framework.

Even more shocking is that the author of this paper is the same top mathematician who just proved the twin prime conjecture in the Annals of Mathematics a few days ago. Everyone, get ready for an earthquake in the field of deep learning."

Once this blog post and tweet were published...

In less than half an hour, it caused a huge stir in European and American computer academic circles.

Countless computer professors teaching at universities, researchers working in early AI labs of tech companies, and PhD and Master's students began leaving comments under Charles's blog.

"Charles, are you sure you're not joking? The problem with asynchronous frameworks was solved by someone doing pure mathematics?"

"This is crazy! Is the crossover really that ridiculous? A mathematician is coming to teach us how to write low-level architecture code?"

"Wake up, person above. Strictly speaking, Zhao Yang is also a computer expert. He published an image processing paper in TIP before, so he's not exactly a complete outsider now!"

"If the news is true, a model built on mathematical intuition is definitely much more advanced than us banging our heads against communication protocols over here."

"Can't wait to see the preprint! There's no suspense about this year's NIPS Best Paper."

With the borderless nature of internet transmission...

A few hours later, due to the time difference, it became daytime in Dragon Kingdom.

Some Dragon Kingdom international students and domestic IT professionals, accustomed to browsing the latest academic trends on European and American technical forums and Twitter, saw Charles's update.

They quickly took screenshots of the English text, attached accurate Chinese translations, and posted them directly to some of the largest domestic academic discussion areas and social platforms.

The computer section of the SMTH BBS, the Machine Learning Bar on Baidu Tieba, and Renren.com, which was most commonly used by college students at the time, all saw related reposted threads almost simultaneously.

《Nuclear-Level Breaking News! Senior NIPS Editor Confirms: Zhao Yang Strikes Again, Solving a Core Problem in deep learning!》

《After the twin prime conjecture, God Zhao Makes a Cross-Field Kill! One Paper Shakes Silicon Valley!》

The matter of Zhao Yang's mysterious paper quickly sparked extremely widespread discussion domestically. The heat of the discussion was even higher than that of the previous mathematics paper, because computer science is closer to the public's actual life.

Those within the profession were discussing what mathematical model the paper used to solve the stale gradient.

Meanwhile, those outside the profession focused all their attention on Zhao Yang himself.

Everyone felt extremely curious, even horrified.

"My god! Peking University and Beijing Forestry University only just held their press conference the day before yesterday. It's only been two days, and he's published another top computer conference paper?"

"Wait, can someone explain what level NIPS is? Is it more impressive than the Annals of Mathematics?"

"Replying to the person above: Different fields can't be compared directly. But in the fields of artificial intelligence and machine learning, NIPS is the undisputed altar, an existence worshipped by all big tech companies and top universities."

"The speed of publishing papers is just too exaggerated, isn't it? Is he really an eighteen-year-old freshman? Does he not need to sleep?"

"I seriously suspect there's a massive national-level secret team behind Zhao Yang. How can one person move without any bottlenecks between mathematics, image processing, and distributed computing—disciplines that aren't related at all? Human energy is limited!"

"I also think there's a team. No matter how high the IQ, writing code, running data, and formatting—these all require time to pile up!"

On the domestic internet, speculations about whether Zhao Yang had a team of ghostwriters behind the scenes were rampant.

In reality, the influence of this matter was still spreading through the networks within the system.

Peking University, Office of the Dean of the School of Mathematics.

Academician Wang Wenyuan had just finished the morning teaching and research meeting and returned to his office to sit down with his thermos.

The landline on his desk rang. The caller was an old acquaintance from the Computer Science Department of Tsinghua University, who was also a leader in early neural network research in the country.

"Hello, Old Li? How do you have time to call me today?" Wang Wenyuan took a sip of water.

"Old Wang, are you familiar with that Zhao Yang from your Peking University who just registered for a joint PhD program?" Old Li's tone on the other end was very urgent.

"I'm familiar, he just gave a two-hour defense lecture here the day before yesterday. He's a genuine genius. Why?" Wang Wenyuan was a bit puzzled; why was a computer guy suddenly asking about someone in number theory?

"He submitted a paper on an asynchronous distributed framework to NIPS. Foreign editors are already going crazy about it on Twitter, saying he solved a low-level computing power bottleneck that's bothered us for years. Do you know about this?" Old Li asked loudly over the phone.

Wang Wenyuan's hand holding the thermos paused.

"What did you say? Submitted to NIPS? Distributed architecture?" Wang Wenyuan's eyes widened slightly, his brain unable to process it for a moment.

He certainly knew Zhao Yang's computer skills were good; that TIP paper was proof. But that was only an image processing algorithm leaning towards the application layer. What Old Li was talking about now, distributed architecture, was the most hardcore and low-level stuff in computer science.

"This kid was still deriving variational weight functions on the blackboard for me the day before yesterday, and he wrote a top-tier computer architecture paper after going back yesterday? Where did he get so much time?"

Wang Wenyuan asked this question directly in his mind.

Although everyone had accepted that Zhao Yang was a once-in-a-millennium peerless genius, even with a high IQ, it was impossible to seamlessly transition from a high-intensity bout of mathematical brain drain into extreme research in another completely different field.

Is he even human?

"Old Li, don't worry yet. I'll hang up and verify this immediately."

Wang Wenyuan put down the receiver and immediately took out his cell phone.

At the same time...

Beijing Forestry University, Administration Building.

Zhou Limin Zhou Limin also received a private call from an official in the Information Science Department of the Ministry of Education, inquiring about the situation just as they had with Wang Wenyuan.

After hearing the call, Zhou Limin was completely stunned.