Because it was not early, Ding Aiying didn't have time to buy food, so she ordered a hearty takeaway.

My son finally came back, and he couldn't treat him badly.

When Liu Suian and his parents finished dinner happily, it was already past nine o'clock.

Liu Suian excused himself to be busy with the project, returned to his room, and closed the door.

The next communication with Bai Ze, he didn't want his parents to know.

First of all, Bai Ze's origin is unknown after all, and it is difficult to explain to his parents, and his parents can't give any advice after they know it, which can only increase their worries.

Secondly, Bai Ze is bound to be very involved, if something goes wrong in the future, it will be Liu Suian himself who has an accident, and if the uninformed parents are lucky, they may not be implicated.

Let's hope so.

Haikou has already boasted, and then he has to think about how to successfully get 500,000 in a short period of time.

When Liu Sui'an closed the door, Bai Ze's voice appeared at the right time.

"You panicked and said that the project was about to be divided into 500,000 yuan, but in fact, you were preparing to participate in the AI painting competition?"

"That's right, it's the most appropriate way to make money right now. You're going to help me, the cutest and kindest little Shirasawa in the world? ”

"Cold knowledge, even if it's not so pompous and pretentious, Bai Ze will help you."

"Love you."

Liu Suian was polite to Bai Ze, and his hands were not idle, he took out his laptop from the suitcase and put it on the desk, plugged in the power adapter and connected it to the power supply.

Before lying to his parents, he thought of many other ways to make money, such as borrowing money from others.

borrowed money from Zhang Boheng or the super-rich second-generation Han Qiming, whether it was good to say otherwise, even if they borrowed it, they would have to pay it back sooner or later, it was nothing more than tearing down the east wall to make up for the west wall.

What else does he have besides Goldfinger Bai Ze?

I only have recently completed my graduation project and thesis, but this kind of thing is too inconvenient to cash in.

If you want to make money about neural networks, you need to apply code to the product and then make a profit, which is the kind of engineering project that Zhang Shasha does.

Specifically, it can be the most common intelligent face recognition system, or the automatic detection system of defective products in the factory, etc., this kind of project needs to deploy the neural network in the scene that can be applied in the field, referred to as the neural network landing.

The topic of Liu Sui'an is an academic project, which can only be said to have great potential, and has not reached the level of landing.

In the future, after publishing the paper, even if he is not open source, others can reproduce his code according to the explanation in the paper.

If you want to get quick money, the only possibility is not to publish papers and sell the code and algorithms to big Internet manufacturers, but this is too troublesome, not to mention that the papers are gone, and there may not be any manufacturers to buy them.

As for applying for a patent and then transferring the patent, it is even more unreliable, this kind of algorithm patent is different from invention patent, and it is difficult to make money.

After thinking about it, Liu Suian thought of the advertisement he saw on the programmer's website yesterday, the company is holding an AI painting competition, and the champion can get 500,000 yuan after tax.

In the field of neural networks in China, Jixin is the strongest existence.

This cannot be denied by all relevant practitioners or researchers.

In terms of products, Jixin has launched a large number of intelligent products, such as intelligent robots, driverless cars, intelligent security inspection systems and so on.

On the academic side, Jixin has a top-notch artificial intelligence laboratory, which has a special name called Jueqi Lab, which produces an average of four CVPR papers related to neural networks every year.

The original neural network model used in Liu Sui'an's graduation project is the handiwork of the AI Neural Network Lab.

For such giants, holding activities such as AI painting competitions can further build their own image of hard-core technology companies, promote the development of domestic neural network technology, and break the suspicion of the monopoly of Jixin companies in public opinion.

The computer boots up.

Liu Suian's fingers were like a jumping elf, tapping quickly on the keyboard, and he quickly found the official website of this AI painting competition.

It's the hand speed he's honed over the years of writing code.

When you open the official website, you will be greeted by a beautifully crafted homepage with a promotional video voiced by AI in the center.

"Every small step forward in AI represents a step closer to a better life. Hello programmers, this is Jixin Artificial Intelligence Co., Ltd., the organizer of the AI Painting Competition.

With the advent of AI painting, you can simply type in a description, or even just a few keywords, and you can get a picture in a few seconds, which is cool, isn't it?

Nowadays, AI painting technology is in the ascendant, and the answers given by AI are still not perfect, and there is a lot of room for improvement.

In order to promote the development of AI painting, Jixin Artificial Intelligence Co., Ltd. specially launched the AI painting competition and put 1.5 million yuan in the prize pool, of which the champion can exclusively enjoy 500,000 yuan after tax.

If you're tempted, don't hesitate to bring your neural network to the contest! ”

There is no key information in the promotional video, Liu Suian fast-forwarded and after watching the promotional video, he controlled the pulley of the mouse and directly slipped the column of the rules of the competition, and read it carefully.

"Preliminary round: October 24 to November 6 at 23:00. During this period, participants can upload their own neural networks to the official website.

After the preliminary round, the upload portal will be closed, and the organizer will select the best ten neural networks to participate in the finals according to the evaluation rules.

The final will start at 7 p.m. on November 7 and will be livestreamed, with the four judges and the audience scoring the drawings of the 10 neural networks, ranking them based on the results of the preliminary round, and awarding prizes according to the rankings. ”

The preliminary round is due before the 7th, and if you write it from scratch, you won't have enough time.

However, with Bai Ze's efficiency, this time is more than enough, and it is not a problem.

Liu Suian then looked down.

"Evaluation process: Participants send the code to the official website in the prescribed format, and the organizer uses the same dataset to train all neural networks, and then tests all the participating neural networks through the input and output interfaces of the code."

A dataset refers to a huge number of pictures that a neural network has to learn.

Like humans, if a neural network wants to draw, it needs to learn knowledge in advance, which is the training of neural networks.

This process has an interesting name, alchemy.

That's right, the training of neural networks is the same as alchemy, for most people, it is difficult to understand what is going on in the alchemy furnace, just add something to it.

If you think about it this way, a competition like the AI Painting Contest is like an alchemy conference, where the contestants prepare things like alchemy furnaces, and then the organizers use the same raw materials to start the training.

Further down is the evaluation process of the alchemy conference.

The specific operation is to enter the same text, score the performance of the neural network through the evaluation rules, repeat ten times, and take the average.

Evaluation rules: the quality of the painting (through different official neural networks, for the originality, aesthetics and picture quality of each painting, etc., the weight accounts for 70%), the drawing speed (the weight accounts for 20%), and the memory occupied by the neural network during the operation (the weight accounts for 10%).

Note: Participants should pay strict attention to the originality of the code, if the audit finds plagiarism, the results will be invalid. Entrants are not allowed to open source their code (publicly distributed) until the results are announced. At the same time, please rest assured that the organizer will not disclose or covet the participant's code. ”

Tight-fitting.

After reading the whole article, Liu Suian made such an exclamation.

It is worthy of the competition founded by Jixin Company, and the rules are very fair and detailed.

For the contestants, the process of participating in the competition is equivalent to putting their own neural network in a black box that is not public, with text input and image output.

Since it's not public, entrants don't need to worry about others seeing their code.

You know, some of the contestants are not only for prize money, but also to make a name for themselves through the competition.

As the promotional video says, AI painting is in the ascendant and the market is huge.

Major companies coveting the interests are gearing up to rely on this competition to make a name for themselves and seize market share.

When a user chooses an AI painting software, he must choose the strongest one, and if the core code of the product is seen by others in the future, what else will he play.

Of course, in order to verify the originality of the code, the internal personnel of the company will verify the code.

But few people are worried that someone from Robotics will leak the code or take it for their own use because they don't need it.

This is the authority and prestige that Jixin has accumulated over the years.

After understanding the rules, Liu Suian found the registration entrance and registered himself as a contestant.

The ID is, after years of peace.

Liu Suian took the name of the contestant, filled in other information in turn, and bound the bank account.

If you can get good results, the company will directly send money through the bound bank account, without the need for the contestant to come to pick it up.

It's very human.

Silently watching Liu Suian register an account, Bai Ze asked: "What algorithm do you want to use to participate in the competition, neural network or general?" ”

"If I only use the neural network algorithms that exist on the planet, can I be guaranteed to win the championship?"

Liu Suian knew that Bai Ze would definitely be inclined to let him choose the Rim, that is, a mysterious and unknown artificial intelligence algorithm that had never appeared on the earth today.

Because promoting the general is her mission.

Bai Ze was silent for a moment, and then gave a firm answer: "Yes." ”

Hearing Bai Ze's answer, Liu Suian secretly felt funny, she didn't know if she couldn't say no, she could only choose to be good?

This stupid artificial intelligence can't even tell lies.

"Then let's use the same language." Liu Suian said quietly.

"Good. Bai Ze first checked the ...... that is generally used now Hey, what do you say? ”

Bai Ze reacted, his tone full of disbelief.