The venue that Sydney gave to the IMAGENET-2013 competition this time was not very big. It was quite impressive to have one or two hundred people sitting there.
As speakers who needed to take the stage, Meng Fanqi and Han Ci were arranged to sit in the first row. Tang Juan's name was in the participating team, so Zhan Guang also sat in the front, and the others had to sit in the back. There are no seats because there are so many people.
The organizers did not expect that the number of visitors this time would far exceed the expected number, so even if they added several rows of seats, there was nothing they could do. Those who came late would have to stand.
Meng Fanqi looked back and saw that the other three people were standing at the back, all smiling and waving to him.
While Meng Fanqi was busy negotiating with Google and studying the contract these days, a few of them walked around the venue of the International Computer Vision Conference.
After all, the conference tickets alone cost hundreds of dollars, so they couldn’t waste it. After walking around for a while, several people felt deeply.
It turns out that there are already so many universities, companies or institutions starting research in the field of AI, with so many directions and types.
And these scholars and industry giants are busy today, standing in this room to watch the technical speeches of their classmates.
Although in the end it had nothing to do with them, the four of them felt more or less happy.
Meng Fanqi glanced over and saw many familiar faces, including Li Feifei, a well-known professor who had become famous for hosting the competition.
There are also future talents like Jia Yangqing who are still studying and competing and are currently unknown.
Everyone is looking forward to today's technical demonstration. What kind of magic can make thousands of categories on tens of millions of images surpass the human level.
But now Meng Fanqi's main energy is no longer on today's display.
Just a few months ago, he worked hard to win this event and gain the attention of the world. But at this time, he had signed contracts with two technology giants and no longer took this event seriously.
Maybe this is the reason why Hinton, Alex and others did not participate this year. Although before becoming famous, they relied heavily on such platforms and opportunities.
But after having a real career, such competitions seem a bit insignificant.
Only Han Ci beside her was still nervous. There was a room full of scholars and professors waiting to hear her thoughts.
Meng Fanqi's outstanding results made the results and algorithms of other teams in this tournament worth mentioning, and no one cared about them anymore.
This somewhat frustrated other contestants who had been working hard for several months.
But compared to the expectation and curiosity about the new algorithm, this little emotion is nothing.
It may also be clear that after Meng Fanqi finishes introducing his algorithm, the speeches of other teams may be ignored.
The organizers deliberately placed his order at the end.
Otherwise, it would be a bit embarrassing if he was the first one to go up and half of the people in the room left after speaking, or if Meng Fanqi was grabbed for discussion during the follow-up speeches of others.
The content of the speeches by everyone on the stage was nothing more than a reproduction of Alex's algorithm last year, with a few small changes of their own.
A variety of modified models are violently piled together, without many new ideas in it.
Glancing at Alex next to him, Meng Fanqi said to himself, now even Alex has to ask me for advice on network structure design.
It is true that listening to such content is useless. Meng Fanqi did not continue to pay attention to the performances of these teams, but looked at Google's financial reports on the computer.
If you want to extract the most money from Google, you must have a very good understanding of Google's advertising revenue structure and specific circumstances, where its money comes from, what the proportions are, and what you can do.
Although he understands some of Google's classic recommendation and advertising algorithms, Meng Fanqi has never read Google's financial report himself and has a very general understanding of Google's revenue structure.
We only know that search and advertising revenue account for 80% of Google’s total revenue.
Judging from the financial reports of the past few years, Google is undoubtedly the number one player in advertising in the world.
It accounts for approximately 30 percent of all advertising revenue worldwide.
For comparison, Meng Fanqi roughly understands that around 2020, the advertising share of the entire Ali system will be about 8-9%.
That is to say, one Google can basically match the advertising market share of four Alis.
Among all Google’s search and advertising revenue, search accounts for the absolute majority, accounting for about 70%.
In fact, Google also does bidding ranking, which is the same as whiteness.
It's just that Google's approach is more diversified and more covert, and advertisers can choose different bidding strategies.
Google allows and has the technical strength to deliver ads and search results quite accurately.
Specific to your geographical location, commonly used languages, personal interests, etc.
Therefore, his advertisements are far more hidden than whiteness, and are not forced on you, although there is actually no difference in nature.
"In other words, my first priority is to complete the optimization of the search recommendation algorithm, followed by the advertising strategy."
As for YouTube videos, they only account for one-tenth of total advertising revenue. Meng Fanqi frowned slightly when he saw this.
All advertisers love video. As the largest video website in the world, YouTube’s revenue from advertising is far lower than that from search.
"It seems that I still underestimated the power of text information." Meng Fanqi thought about it carefully. At this stage, Google lacks the ability to identify information in videos.
Therefore, the preferences of video users are far less accurate than the preferences of web users.
"I need to focus on investing in this direction." Search is Google's trump card income direction, and its own harvest is divided according to the additional income.
Although the proportion of search revenue in Google is seven times that of video, this part has nothing to do with his commission. He can only take part of the revenue from additional optimization.
Meng Fanqi believes that the space for video advertising may not be smaller than that of search, and it is his best visual field. On the contrary, pure advertising can be ranked last.
He was immersed in studying, and before he knew it, it was his turn to go on stage to demonstrate.
People who didn't know about it later thought that the amazing Meng was still actively studying until he arrived at the venue, staring at the computer and studying technology. No wonder he could produce so many amazing results.
Only Jeff, who was sitting next to him, could see clearly that this kid was basically studying which technology could divide the money more.
This is the first time in my life that I have chosen technology based on financial reports.
"Okay, I think everyone can't wait. We all know who is the real reason why the venue is so crowded today!" After the last team left, the host came up and said with a smile, "Let us welcome The rising star in the field of vision, Meng! Explain to us his huge breakthrough in the field of image recognition!"
