Search Google for "What is the world's fastest bird?" It will tell you: "You cruising. According to the YouTube video, you can fly at a speed of up to 389 kilometers per hour."
This is the correct answer, but this result is not based on Google's cool database. When you enter a question, Google search engine will accurately capture the information you need in a YouTube video, such as extracting the fastest bird from the video "The world's fastest five birds." The information, but also automatically shields the other four birds.
With the deep neural network technology, Google successfully achieved this, which is a kind of artificial intelligence. This breakthrough will not only reshape the face of Google's search engine, but also directly affect the entire company's future technological development. At the same time, it will exert pressure on other Internet giants such as Facebook and Microsoft.
Deep neural network is a pattern recognition system that can learn how to complete specific instructions by analyzing huge data. The above example shows that it has mastered how to intercept long sentences or paragraphs in related web pages from the Internet to answer your questions and accurately find the focus of the questions.
In other words, "Sentence compression algorithms" are the core of this technology. This semantic interpretation is simple for humans, but machine learning has always been difficult. Deep learning is currently developing rapidly in terms of natural language understanding, that is, how to understand and respond to human language capabilities.
Google Academy product manager David Orr said: "The implementation of this task must use neural network algorithms, or to say so, this is the only way we know so far. We must of course have to use our hands. There are all the most advanced technologies available."
Not to mention you need to have a bunch of smart brains. These smart brains all have PhDs. Google has such a strong linguistics doctoral team called Pygmalion. They are training Google on how to use neural network algorithms. First, let the machine repeatedly observe how humans extract relevant information from an article. This difficult learning process illustrates the power and limitations of deep learning. To train this artificial intelligence system, you have to personally come to personally choose a huge database. This database is not simple and not cheap. Moreover, now that the machine can learn on its own, it can be too early.
"Golden Data" and "Silver Data"
In order to train the AI ​​ability of Google artificial intelligence, the team let the machine read a lot of old news, through the news headline can learn how to sum up a long article. For now, Google relies heavily on this team of Ph.D. linguist experts. They can not only demonstrate compressed sentences, but also classify the language components into different parts of speech to help the neural network learn the mysteries of human language. About 100 such linguists around the world are co-creating these so-called “golden data†inventories. Those old news reports are “silver data†and their use is of course great because they are large. But gold data is the most essential.
Linne Ha, head of the Pygmalion team, revealed that in recent years the team will need more and more linguists.
This artificially assisted artificial intelligence system is called "supervised learning". This is how the current neural network works. Sometimes the company can crowd out this business, and more often it is already happening spontaneously. For example, there are countless people on the Internet that add pictures of cats to the label of “Demiâ€. This makes it extremely easy for a neural network to learn to recognize Demi. But more often than not, researchers have to manually add tags themselves.
Chris Nicholson, founder of deep learning startup Skymind, pointed out that in the long run, this manual method of manual tagging is not feasible. “The future will definitely not be like this. What a boring job it is! I can hardly imagine a job that is more boring than this.†The bigger problem is that if the linguist on the valley singer is not If you include all the languages ​​of the world, then this system will not work in other language environments. Currently the Google team covers almost 20 to 30 languages, which means that you can only expect to develop more automated artificial intelligence in the future, such as "unsupervised learning".
In this way, the machine can learn on its own, and it is no longer necessary to manually tag the data manually. "Unsupervised learning" can analyze and learn through the massive information on the Internet or other channels. Google, Facebook, and OpenAI, founded by Elon Musk. These giant companies are all working on this field, but there is still a long way to go before it becomes a reality. Currently, artificial intelligence still requires a linguist like Pygmalion. team.
[This article from the Tencent News client from the media, does not represent the views and positions of Tencent News]
Recommended: Focus on Tencent Digital WeChat official number (ID: qqdigi), take you through all kinds of cool new products, cutting-edge information, one-handed evaluation, video games, interesting live broadcast, as well as a variety of novel play methods. How to use one plus 3T? What is the most feared Drone? What should I do if I drink cold water? There are several iPhone 8 prototypes? The answer is here.
MARSHINE splice sleeve straightener serves to straighten the bent splice sleeve with its specification of LGJ720 or below after the crimping process. This product weighs 12kg, and its maximum straightening force reaches 20kN. The item number is 17211.
Ningbo MARSHINE Power Technology Co., Ltd. is a professional engaged in the development, design and manufacture of power engineering construction equipment and tools.
MARSHINE company produces circuit construction tools, including foundation construction, tower group lap. Wiring structure. Cable construction, cable construction, mobile knife mill, insulated overhead cable and high voltage cable stripper, all kinds of aluminum alloy pull rod, guide rail, grounding device, high strength shackle, ratchet wrench and pointed wrench, double hook tight line device, lifting pulley, nylon wheel and aluminum wheel, punching machine ect.
MARSHINE continues to carry forward the enterprise spirit of "integrity, development, innovation" and strive for the prosperity and development of the electric power industry.
Welcome to contact MARSHINE and reach cooperation, thank you!
Splice Sleeve Straightener,Splicing Sleeve Straightener,Fusion Splice Sleeves
MARSHINE , https://www.puller-tensioner.com