I Have a Little Black Hole

v1 Chapter 347: Difficulties in the research and develop

"You can think of it this way to show that you have already regarded the company as part of your future in your heart! This is very good! But no matter how much investment you invest in, I will come out, and you are not without contributing to the company, some things are not just It can be massive with money!

For example, the company’s current security system, unmanned operating systems for various production lines, etc. are not all developed by you. These things cannot be bought with money. Even if I buy them, I can’t use them with confidence, so you Everyone can see the contribution of the supercomputer department to the company!

But since you want to bring greater benefits to the company, I am very grateful! Helping the police to develop their artificial intelligence system is indeed a good thing with multiple birds, allowing you to see how artificial intelligence should progress from more perspectives, and it also plays a very important role in promoting the upgrade of ‘Origin’!

So it's up to you to win the police contract this time! If you want to do it, you must do your best. Just say what resources you need. There is no problem with how much money you need. You decide how many people you need to recruit! "Lei Tiantang said with a smile.

"Boss, don't worry! Our'origin' is not the simple artificial intelligence as it was when it was first born. Its current intelligent learning system has evolved very well, and it will not be like the so-called artificial intelligence that the police are using. Can only concentrate on the field of video images!

It can now perform self-smart labeling! Now its database is growing at a very fast rate every day! If our supercomputer department needs to help the police develop exclusive artificial intelligence, we really need the boss to continue to invest!

The computing power of "Origin" needs to be further improved, its storage space needs to be further expanded, and the continued recruitment of those computer experts also needs investment! "Wu Wenyue replied with a smile.

As for Wu Wenyue's statement, Lei Tiantang also knows that the artificial intelligence applications currently used by the police are indeed only concentrated in the field of video images! This is determined by social development.

With the development of machine deep learning algorithms, artificial intelligence has developed rapidly in machine vision, fingerprint recognition, face recognition, retina recognition, iris recognition, palmprint recognition, speech recognition, semantic analysis, etc., and has been widely used.

At the application level, the artificial intelligence technology platform mainly focuses on the fields of computer vision, speech recognition and language technology processing, such as: intelligent translation, intelligent education and other human-computer interaction, unmanned vehicles, drones and other autonomous driving, precise push, precise search, etc. Smart application.

The above intelligent applications are the application results of machine deep learning technology. Machine deep learning is divided into four methods: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.

Behind this learning is often inseparable from the labeling of relevant data used for training machine deep learning: polygon labeling, point labeling, line labeling, 2D frame labeling and other image annotations, as well as voice annotation, video annotation, text annotation, etc.

Some people say: how many people there are, how much intelligence is there. The labeling of these training data relies on a large number of people to clean and label the data, and then use it for machine learning and training.

Where there is demand, there is a market. As the demand for machine learning becomes more and more prosperous, more and more related data annotations have spawned many data annotation companies.

The business of these data labeling companies mainly focuses on voice labeling, image labeling, and video labeling. Therefore, related applications are also developing fastest in these directions.

If the application of artificial intelligence is introduced in the process of public security intelligence analysis or the pursuit of suspects, it will inevitably involve the cleaning and labeling of relevant training data, such as case classification, personnel classification, alarm classification, and so on.

If this kind of data is handed over to a general data company for labeling, it involves confidentiality issues and labeling accuracy issues, so it is more difficult to label than ordinary data.

Therefore, the basic work of developing the exclusive artificial intelligence used by the police is not only to collect data, but also to standardize the processing of data to lay a solid foundation for subsequent machine deep learning.

To realize the intelligence research and judgment of artificial intelligence, it is inevitably inseparable from the standardized artificial intelligence research and judgment labeling. The artificial intelligence alarm and handling system used by the police needs to automatically convert the voice alarm into text and extract keywords.

Then the alarm is automatically classified and labeled, and at the same time, it is correlated and compared with the video, image, and other data that are also structured using artificial intelligence, and suspicious persons are found in a very short time and an early warning is generated.

Artificial intelligence has natural advantages in feature extraction and content understanding of video content. The front-end camera has a built-in artificial intelligence chip, which can analyze the video content in real time, detect moving objects, identify the attribute information of people and cars, and transmit it to the back-end artificial intelligence central database for storage through the network.

Utilizing powerful computing power and intelligent analysis capabilities, artificial intelligence can analyze the suspect’s information in real time, give the most likely clue suggestions, and lock the suspect’s trajectory from a few days to a few minutes. The detection saves precious time.

With its powerful interactive ability, it can also communicate with the police in natural language ~lightnovelpub.net~ and truly become an expert assistant to the police.

And if the police want to do their own research and development, the work of labeling these materials can make them collapse, because the workload is simply astronomical. It is necessary to know that criminals can use more criminal methods. The labels required for artificial intelligence to accurately determine the trajectory of criminals may be different every time.

If you rely on manual labeling of these data, it may take thousands of people to work for several years, but the "origin" of Infinite Gravity Group's artificial intelligence has now grown to the point where you can learn to label data by yourself. These tedious tasks As long as the people in the supercomputer department set up the initial conditions, "Origin" can learn the information provided by the police by themselves!

With such an advantage, it is no wonder that Wu Wenyue will have the confidence to win the police's order. The speed of research and development alone can explode several blocks of competitors! Therefore, Lei Tiantang certainly agreed to his funding request!

"I will grant you 1 billion U.S. dollars at a time. The money should be enough to double the performance of Origin! Next time it will expand its performance, I believe that our company's CPU chip can be developed. At that time, replace it with our company's own CPU chip, you can stack the performance of Origin to any extent!" Lei Tiantang said with a smile.