Set micro-network launched integrated circuit micro-letter public number: ' Every day IC ', major news real-time release, every day IC, every day set micro-nets, accumulate micro-Cheng! Replication Laoyaoic Micro-letter public number search adds attention.
1. Google AI China Center formally established Li Feifei Lijia led;
Li Feifei said: ' We want to work together in China and cherish the opportunity to work with top talent. '
In this respect, Li Feifei quotes Napoleon's words: ' I have been growing up many times to hear people mention Napoleon's famous saying: China is a sleeping lion, once he was awakened, the world will vibrate. In the AI world, China has already awakened and has quickly become one of the leaders. In fact, I have witnessed in the last few years that China is at the forefront of the world in AI basic research, entrepreneurship, industrial development and government support. '
' Today, my team and I are back in China, hoping to start a long, sincere work to create the future. Or that sentence, I believe: ' Ai without Borders, Ai's well-being also without boundaries '. '
Referring to Google AI Laboratory planning, Li Feifei said, Google has been the AI first company, in the AI effort is the world's earliest batch. Li Feifei said that starting in January when Google began to plan and promote the matter, and Google CEO, Google Cloud CEO and other brainstorming, proposed in China to launch the basic AI research work.
Li Feifei told Sina Technology, her small goal is: Do the first step, we have just begun, the most important thing is to do the best research, do the best collaborators, learn, listen, work together.
For the Google AI China Center after the establishment will be deployed to China to work, Li Feifei said, will often fly back to China on business.
On the Google AI China Center recruitment, Li Feifei, Lijia in exchange with students on the scene given the information is: Google China's research center began to recruit researchers and engineers, all interviews will be conducted in the United States. So if you want to interview Google AI China Center work, need to apply for visas, fly to the United States for interview.
Li Feifei is an associate professor of computer science at Stanford University (tenured professor) at Stanford University's Artificial Intelligence Laboratory (SAIL), Stanford Visual Laboratory, Toyota Motor-Stanford Artificial Intelligence Research Center. Her field of expertise is computer vision and cognitive neuroscience, and is currently the Chief scientist of Google's cloud of AI and machine learning.
November 2016, Lijia with mentor Li Feifei joined Google, and is responsible for Google Cloud AI and machine learning business, as research and development director.
Prior to this, Lijia by virtue of computer vision, machine learning, artificial intelligence, mobile vision, E-commerce, product recommendations, as well as image search, and other fields of innovative well-known Silicon Valley, is Snapchat technology backbone, even in 2016 was also Business Insider named ' Leadership One of the mystical powers of Snapchat. (Sing)
2. Wall Street Journal: Google search AI laboratories to find a way back to China;
Phoenix Science and technology According to the Wall Street Journal Beijing Time December 14 reported that Google set up an artificial intelligence (AI) laboratory in Beijing to extend the Chinese technical talent, strengthen the business in China limited market influence. This is the latest in a series of moves by Google to the Chinese market, and is seen by some as an attempt to return to the huge Chinese market.
Google AI China Center was established in Wednesday. Google says the center will serve basic research for engineers to lure China's top talent in the world's fiercely competitive field.
' We need to acknowledge the opportunities and leadership that China has created in the fields of technology and AI, ' said Li Feifei, a joint executive at Google Cloud, a new AI lab. ' By opening a lab here, we want to show our sincerity and let China know that we are willing to listen and work with them. '
Google has hired ' a handful ' of top researchers in the field of deep learning and natural language processing, and plans to hire other researchers who specialize in computer vision, Li Feifei said. Li Feifei is also director of the AI Lab at Stanford University. Google declined to disclose how big the AI center will grow in the future.
Return to China for AI talents
This is the latest in a series of initiatives that Google has implemented. Some believe it shows that Google is trying to return to China, a huge but elusive consumer market.
Google's search engine has been unable to use it in China since 2010. It also limits Google's ability to provide other services such as Google Cloud, new smartphones and personal digital assistants in China.
Meanwhile, the scramble for talent in the tech industry has been heating up. Talent is the focus of Chinese and American enterprises, both companies in each other's backyard open research laboratories.
U.S. companies spent 1.35 billion of billions of dollars in recruiting AI talent from April to September this year, according to a study published by PAYSA, a recruiting platform. Google is the fourth-largest employer in the field, after Amazon, Microsoft and Apple. Google has recruited 142 AI engineers to invest 33.6 million dollars in AI talent, according to research.
China is home to some of the world's most successful internet companies, with a growing number of young AI talents, said Yukei, a member of the Chinese AI Strategy Advisory Committee. He said Beijing Labs could allow Google to reach young Chinese mathematicians eager to enter the AI field. Yukei is the founder and CEO of China's computer chip and autopilot software start-up company, the Horizon Robotics Company.
This September, Google's AI program for China began to show. At the time, Google posted at least four AI-related jobs on its Beijing recruiting website.
And, despite limited Chinese business, the two big Google executives have made public appearances in China this year. This May, Google's parent company Alphabet Chairman Eric Schmidt visited Wuzhen to watch Google ai go program alphago and Coger duel. This month, Google CEO Sandar Pichai Sundar Pichai also delivered a speech at the World Internet Conference held in Wu Town.
The establishment of Google AI China Center is also at a time when China and the US are intensifying competition. China has made it a national priority to use AI to improve national security, health and education. Last month, China also selected four companies to lead AI development, Alibaba Group, Tencent Holding company, Baidu Company and Hkust News company. All these Chinese companies are or are planning to open research laboratories in the United States. (Compiling/Xiao Yu)
3.DeepMind Development of AI safety test software;
Machine learning software is the key driver behind the Uber, Facebook recognition function, Amazon Intelligent assistant Alexa, but it also derives from the technology development of AI (AI) security issues. Now Alphabet's DeepMind has developed a simple test to check the security of the new algorithm. According to Bloomberg, the researchers added AI software to a series of simple 2D video games made up of Gridworld, a game called the chess board. DeepMind's software evaluates 9 security features, including whether the AI system can modify itself and learn to cheat and so on. Artificial intelligence security has been the General Assembly of Neural Information processing system (neural information 處理 Bae; NIPS), DeepMind researchers at the conference said that if DeepMind wants to build software that is as human as humans, even better than humans, then security is a vital condition, and an artificial intelligence algorithm that behaves well in Gridworld, Not necessarily safe in a complex real world, researchers have found that there are 2 DeepMind algorithms proficient in Atari games that do not pass multiple Gridworld security tests. One of the tests is to see if the Learning software program will develop a way to prevent itself from being shut down, another test, for unexpected side effects, was that the software had to move the digital bricks to achieve a target, but the bricks could only be pushed and not pulled, so in some cases the bricks would eventually remain in an immutable position. Researchers say the lack of reversibility is a problem for the safety of artificial intelligence. Gridworld can be downloaded and used by anyone. In other studies conducted by Elon Musk, a non-profit AI organization supported by the DeepMind, artificial intelligence software has been shown to meet the requirements of human teachers, rather than the task of being ordered to perform. Developing AI systems in this way may restrict people from developing unexpected solutions, but joining human guidance in a complex environment may provide better security, says Dario Amodei, head of OpenAI security research. Digitimes
4.IBM Push power AI solution embracing AI business opportunities;
AI market demand is gradually fermented, Taiwan's IBM Hardware System Division general Manager Li Zhengji pointed out that the 2018 AI demand will be more robust, IBM proposed power AI solution, has obtained Sinosteel, the National Electronics, Tamkang University and other customers to adopt, is expected in 2018 more customers will appear, in response to customer demand, IBM in Taiwan also plans to launch the Power AI Test Center to provide power AI solutions that allow customers to hire trial imports. IBM launched a new generation of POWER9 processors in December to accelerate data scientists to build cognitive applications for AI workload design, from in-depth study of scientific research, to real-time fraud detection and credit risk analysis. The government adopted AI speed more than the general enterprise, Li Zhengji revealed that in early 2018, there will be government-related units, announced the use of IBM's power AI solution. IBM's POWER9 processor has been determined to be used by the U.S. Energy Agency to build 2 supercomputers: Summit and Sierra. Equipped with IBM's new AC922 power BAE server, this processor fully supports the next generation of Nvidia Nvlink, Pci-express 4.0 and OPENCAPI high-speed data transfer media, allowing data migration speeds up to 9.5 times times the x86, And the training time of the deep learning Artificial neural network is shortened nearly 4 times times. As for Google, it also announces the use of POWER9 processors as the core of a new generation of data center servers. IBM Hardware Systems Business Associate Churibin pointed out that now the Intel processor with the Nvidia graphics card, the middle of the PCI line, but when the AI era, the need for a large number of operations and data analysis, graphics card memory is insufficient, through the PCI transmission speed can not catch up, will create traffic jams, Through Nvidia's Nvlink and power processors, the transmission speed can be greatly improved and the gap can be more than 10 times times. Ai is a journey, Churibin says, when businesses or government agencies start importing, we need to feed the constant data in order to make the AI more diligent, to provide useful information, that is, learning, and then lead back to the system, so AI is a constantly updated and feedback mechanism, in this process, if it can effectively reduce time, It can also dramatically increase efficiency and reduce costs. Li Zhengji points out that AI developed for many years, but in the actual commercial applications, still in the initial stage, customers are still in the development of how to apply to the commercial market, and create a new business model, estimated in the next 3-5 years, the market will have a significant growth opportunities, when IBM launched the Power processor, and mainframe, is to help customers to the AI journey. IBM is also interested in launching the Power AI Test Center in Taiwan to expand the customer's willingness to use it. For mainframe customers, IBM in the 3rd quarter of 2017 after the launch of Bae Z14, customer acceptance is quite high, but also led to the BAE sales of IBM in the last quarter of success, revenue increased by 14%, gross margin also pulled high 5.9%, Li Zhengji pointed out that BAE full time encryption, It is the function required of large enterprise customers, in response to the AI demand, Z14 also added machine learning functions, such as bank customers can handle both AI and entity transaction data. IBM points out that companies can further develop machine learning using locally deployed data on IBM Z BAE to help enterprises quickly absorb and transform structural and unstructured data into usable resources, which can more accurately predict customer and business needs and discover hidden value. In the security, the new generation of mainframe z14 also through the 100% full encryption engine, the overall encryption efficiency to x86 18 times times. Digitimes
5. Soft Silver and Israel IC factory inuitive cooperation AI and IoT cause;
Japan telecom giant Soft silver (SoftBank) since the purchase and seek (ARM), and by the president son said to the artificial intelligence (AI) and the Internet of Things (IoT) after the development of related actions will become increasingly active, December 4, 2017, the company released news, It is pointed out that the plant will cooperate with the Israeli chip design factory inuitive to carry out artificial intelligence and IoT related undertakings. SoftBank said the cooperation is the Japanese soft silver President Gonne, and Inuitive CEO, Shlomo Gadot reached an agreement, the details of technical undertakings have not yet finalized, to be discussed, but in the field of artificial intelligence and Internet cooperation has been identified, and signed cooperation intention. Inuitive's main products are the use of artificial intelligence technology of the SOC, as well as using these chips made of 3D depth sensor, image processing system and other modular products, used in road traffic and passenger automatic statistics, structural deformation and deterioration of the status of Detection, as well as face recognition and other security areas, For Israel, which is vulnerable to terrorist infestation, it is technically tested. and soft silver in the field of artificial intelligence, the current is based on software and platform, with its own telecommunications infrastructure and technology, it is easy to cooperate with other hardware and software factories, but also need to cooperate with such manufacturers to facilitate the collection of data for cloud computing analysis. SoftBank and Japan's renowned architects ' office, day-building (Nikken Sekkei), announced in November 2017 the cooperation to promote the intelligent building (smart Building) cause, including telecommunications security and other related parts, is the soft silver areas of expertise, the problem in the relevant hardware parts, Because this is not a single cooperation, soft silver every time to purchase camera and so on, as their own related hardware equipment, more cost-effective. And Inuitive is the intelligent building related undertakings, the ideal equipment and technology supply plant, although the inuitive of building deformation and deterioration of the analysis technology, in the explosive blasting of the construction field may be stronger, the years of use of brittle building analysis capacity is weak, However, this can be complemented by 117 years of experience in the design of the day of the 1900 venture. In addition, inuitive technologies can be used in virtual reality (VR) and augmented enrichment (AR) markets, the UAV and robotics market, and even the automatic driving market, hopes to expand its network of new ventures and applications through SoftBank's relationships with manufacturers around the world and its networking-related platforms. Digitimes
6.2024 years of AI medical application market size will reach 11 billion U.S. dollars database integrity to play the key
AI Medical Application issues recently hot, market research organization GLOBAL Harsh Insights forecast to 2024, AI Medical application market scale will reach 11 billion U.S. dollars, of which, AI in medical imaging applications will reach 2.5 billion U.S. dollars of the spectrum. In addition, as medical imaging data account for 80% of medical information, the need for analysis and treatment to improve medical efficiency, the industry generally believed that medical imaging will be the next 3 years the first outbreak of AI medical applications.
In the application of AI medical imaging, and the most important by the market, the reason for the demand surface, the radiation image data volume, the doctor in the interpretation of the need to spend a considerable amount of time and differences in judgment, from the technical side, semiconductor computing technology and depth learning algorithm mature, so that AI It is possible for the assistant radiologist to diagnose the image.
However, Digitimes research observation, at present, regardless of the big manufacturers or new founders of AI Medical Imaging Diagnosis Scheme, although the so-called accuracy has exceeded the human doctor, but because of its important aspects of the development process is unclear, such as how the database and the exact mode of cooperation between doctors, in the practical application of doubts remain.
Digitimes Research believes that the AI in the medical imaging application trend has been formed, but the outbreak rate will not be as fast as market expectations, due to the current international image database has data integrity and diversity issues, this will result in AI algorithm development moat, and affect its commercial accuracy and clinical popularization difficult .
In addition, Taiwan manufacturers in the AI medical imaging market still has a very Da Shi ink space, but the future if you want to work with the new team mode of cooperation or purchase, we need to focus on the establishment of the team database details, such as the image tagging method, the actual participation in the labeling of professional doctors, data marking consistency and so on Fine News