The annual Google I/O Developer Conference is held in Mountain View, California, USA. As one of the most prestigious developer conferences, this year's conference is still very "hot". At the conference, Google brought a number of new technologies to the eye. When the conference was in progress, a restaurant at the show suddenly caught fire. The Google I/O Developer Conference was really too hot, even the restaurant. They are all on fire.
To be on the safe side, Google suspended the conference and stopped attendees from entering the venue, while evacuating journalists from the media lounge.
The Google i/o conference is indeed very hot, so the most fire is artificial intelligence.
A series of new products and services based on artificial intelligence technology were released at the Google Developers Conference, reflecting Google’s overall development strategy from “mobile first†to “artificial intelligence firstâ€.
The field of application of Google artificial intelligence has been extended to Google for jobs, cancer, and more. Artificial intelligence can be divided into a base layer, a platform layer, and an application layer. The base layer includes basic hardware such as AI chips. Currently, the underlying technology business of overseas technology giants is continuing to break out. NVI's 2017 Q1 revenue increased by 48%, profits increased by 159%, and Google also released the second generation TPU. The maturity and outbreak of the base layer lays the foundation for the upper layer application. As the platform layer algorithm matures, the profit of the AI ​​industry will be transmitted from the base layer to the application layer.
Google clarifies that the company's overall strategy has shifted from the previous "Mobile first" to "AIfirst." The company lays out artificial intelligence from the whole industry chain of “underlying hardware + operating system + core algorithm + upper layer applicationâ€. Cloud TPU, the second generation of TPU released by Google, is a high-performance processor specially designed for AI computing. It optimizes training and reasoning on the first generation. TensorFlow is a deep learning open source tool launched by Google. The optimized TensorFlowLite allows more developers to run deep learning models on Android phones. Improved chip performance and optimization of development tools will lay a good foundation for the establishment of an artificial intelligence ecosystem.
Google CEO Sundar Pichi said in a speech at the conference that computing is evolving again, moving from "mobile priority" to "artificial intelligence first." Google is rethinking all its products in order to apply artificial intelligence and machine learning technology to Google's business as much as possible to solve practical problems. Google is also building an artificial intelligence-first data center.
Pei Chai said that Google has incorporated all artificial intelligence projects into the newly established platform Google. In ai, accelerate the progress of the company's R&D personnel in the field of artificial intelligence. In addition, Google. Ai will also promote collaboration between Google researchers and external scientists and developers to address issues.
In terms of artificial intelligence related products and applications, Google announced that it will launch a "Google Lens" function based on machine vision, which can identify and analyze the scene around the user through the mobile phone camera, and display relevant information on the screen of the mobile phone to help users based on this information. Make a decision. For example, if you point the camera at the signboard of a restaurant, the restaurant's ratings, menus, and ordering options will be displayed on the phone screen.
Google will also integrate the "Google Lens" with Google Voice Assistant, so that the voice assistant has added a pair of eyes, that is, using the mobile phone camera to "see" the user's environment and act accordingly. For example, if the user points the camera at the entrance of the concert venue, the phone screen will display the concert information, and the user can purchase the ticket through the voice assistant.
In addition, Google also announced the extension of the voice assistant function from its Android operating system to Apple's iOS operating system. This means that Google Voice Assistant will appear as a separate app on Apple phones and tablets.
Why "automated AI system" makes Google CEO so excitedThe industry media "MIT Technology Review" website published an article on the importance of the Google AutoML project. This "automated AI system" can help develop AI software, solve a thorny problem in AI software design, and accelerate the process of computer intelligence. The following is the original content:
Companies in many industries are eager to take advantage of the latest artificial intelligence technology, so there is a shortage of machine learning experts. Sundar Pichai, Google's chief executive, said one way to solve the shortage of talent is to let machine learning software help develop machine learning software.
Sandal Pichai at Google’s annual developer conference
At this year's Google Developers Conference, Pichay introduced a project called AutoML, which was launched by Google Brain, the company's artificial intelligence research team. Designing machine learning software for specific tasks has some tricky parts, and the team says that their learning algorithms can automate one of the toughest parts. In some cases, this automated system is designed to rival the best work of human experts.
“This is a very exciting development,†Pichay said in an e-mail. “It can accelerate the development of the entire field and help us solve some of the most challenging problems we face today.â€
Picay hopes that the AutoML project will expand the number of machine learning developers, because with AutoML, developers don't need to have as much expertise and can take advantage of the power of machine learning. Google originally intended to position the company's cloud computing services as the best place to build and host machine learning, so the introduction of AutoML is in line with this policy. The company is trying to attract new customers in the enterprise cloud computing market, which currently lags behind Amazon and Microsoft in this space.
AutoML is designed to make it easier for everyone to use a technology called "deep learning," and Google and other companies use "deep learning" to enhance speech and image recognition, translation, and robotics.
"Deep learning" teaches software to become smart by propagating data into artificial neural networks. Artificial neural networks consist of a number of mathematical layers that are critical to choosing the right architecture. But it is not easy to do this. Quoc Le, a machine learning researcher at Google's AutoML project, said: "We chose the architecture through intuition."
Last month, Quoc Le and colleague Barret Zoph introduced the results of an experiment. In this experiment, they let the machine learning system find an optimal architecture to let the software learn how to solve language and image recognition tasks.
In the task of involving images, this system can match the best architecture designed by human experts. In terms of language tasks, this system is superior to human experts.
Perhaps more importantly, it proposes an architecture that researchers have not considered before, but are well suited to these tasks. “In a sense, it found something we didn’t know,†Quoc Le said. "This is amazing."
The concept of learning "better learning" software is not new. But like many concepts in the field of artificial intelligence, the power of deep learning opens the door to new advances. Another AI research department, DeepMind, and OpenAI, a non-profit organization founded by Elon Musk, are exploring related concepts.
Are their efforts making them unemployed? When I heard this question, both Quoc Le and Zoph laughed. For now, the technology is too expensive to be widely used—the two experiments used 800 powerful graphics processors and worked for a few weeks, causing the electricity bill to soar. This kind of research belongs to the nature of the stone, and few companies can afford it.
But Google has this resource, and now the company has expanded its AutoML project team size, with tasks including reducing the resource intensity of the project. Quoc Le believes this may help improve the accuracy of video or speech recognition, and may even make progress on a more difficult problem: let the software learn without human explicit guidance.
Residual Current Circuit Breaker
RCCB named Residual Current Circuit Breaker. When there is human electricity shock or if the leakage current of the line exceeds the prescribed value, Residual current circuit breaker/RCCB(without over-current protection) will cut off the power rapidly to protect human safety and prevent the accident due to the current leakage. The rccb switch which made from Korlen electric can be used as infrequent changeover of the line in normal situation.
Korlen electric ---- the rccb switch manufacturer,produces types of Residual Current Circuit Breaker. It is applicable to industrial site, commercial site, tall building and civil house.
Residual Current Circuit Breaker,Ac Residual Current Circuit Breaker,Miniature Residual Current Circuit Breaker,Residual Current Electrical Circuit Breaker
Wenzhou Korlen Electric Appliances Co., Ltd. , https://www.zjthermalrelay.com