Introduction: In the past 2020, the global outbreak of the new crown epidemic has not stopped the development of artificial intelligence (AI): Tesla's fully autonomous vehicles are issued in limited quantities, Baidu's global live broadcast of "fully autonomous driving" technology, brain-computer interface The successful application of BCI technology (BCI) in pigs, as well as the decoding of mRNA structures by quantum computing to help vaccine research and development...The results of research and development are numerous.
Recently, many technology giants such as Baidu and Google have released a summary of AI development in 2020. On December 30, 2020, Baidu bid farewell to 2020 with a 4D essay "Baidu AI 2020". On January 12, 2021, Google released the 4D content "Google Research: Looking Back at 2020, and Forward to 2021" to say hi in the new year. Combining these summary articles with important technology events throughout the year, such as Microsoft Build and Amazon re:Invent, we can see several key areas of focus for technology giants in 2020: AI technology supports anti-epidemic, continuous breakthroughs in AI basic technology, Continuously increase the integration of cutting-edge technologies such as quantum computing autonomous driving, cloud computing and AI.
1. AI technology supports anti-epidemic, and life science has become a hot spot for AI applications
Up to now, the global number of deaths from the new coronavirus has reached 2.06 million. Virus testing and vaccine research and development are racing against time, and every breakthrough in achievement may save tens of thousands of lives. However, understanding the molecular structure of viruses, especially viral proteins and RNA, is an extremely complicated process. In other words , whoever has the ability to win over the virus will lead the way in resource allocation and social governance.
At the beginning of 2020, the domestic new crown epidemic is coming fiercely. Baidu's response is the most astute. As early as January, Baidu opened up the linear time algorithm Linearfold, which can shorten the secondary structure prediction of the virus genome from 55 minutes to 27 seconds, which is 120 times faster. In addition, Baidu launched the world’s first mRNA vaccine gene sequence design algorithm LinearDesign, which can greatly improve the stability and protein expression level of vaccine design within 16 minutes, thereby effectively solving the most important stability problem in the development of mRNA vaccines and accelerating vaccines R&D speed.
In the United States, Google’s AlphaFold welcomes breakthrough progress and is expected to apply the results to vaccine development. As an artificial intelligence program developed by Google's DeepMind, AlphaFold can predict protein structure through deep learning. On November 30, 2020, AlphaFold2 won the championship at the Protein Structure Prediction and Evaluation Conference (CASP), and its prediction accuracy reached a new high.
The above results mean that AI is expected to greatly improve the accuracy and efficiency of macromolecular structure prediction. The combination of AI, biological computing, virus research, vaccine research and other basic research fields in life sciences will usher in an explosion.
The predicted value of the team with the highest accuracy in each CASP conference since 2006, and the AlphaFold team has reached a new high in recent years.
The picture comes from the network
AlphaFold high-precision protein structure prediction, the picture comes from the network
In addition, technology giants have successively implemented many AI applications to help governments, communities, and people fight the epidemic. Baidu AI temperature measurement appears in public places such as subways and high-speed trains to solve the problem of body temperature detection when wearing masks and hats; the intelligent outbound platform calls out 1,500 calls per second to assist in health checks and notifications; AI inquiries are called nearly 10,000 times a day, and robot More than 30 hospitals and institutions have implemented war epidemic solutions.
Under the epidemic last year, Google worked with Apple for the first time and jointly launched an Exposure Notification System (ENS) that supports Bluetooth privacy protection technology to promptly notify users to avoid contact with people infected with the new crown. At the same time, Google is also using users’ anonymous search data to predict the spread of the new coronavirus and data related to symptom searches.
In January 2020, Microsoft announced the launch of the AI for Health project, focusing on helping researchers who are on the front line of new coronavirus pneumonia research. Subsequently, at the Microsoft Build conference, Microsoft created and launched a "medical cloud" on the basis of existing cloud services related to the medical industry. This specialized industry cloud is the first time that Microsoft has made it.
2. Continuous breakthroughs in AI basic technology
In terms of AI basic technology, Baidu Brain 6.0's core technology has broken through "knowledge-enhanced cross-modal deep semantic understanding" to understand complex scenes in the real world. For example, the visual language model ERNIE-ViL refreshed 5 mission records and topped the authoritative list VCR, and the ultra-large-scale open domain dialogue generation network PLATO-2 reached a new height in content richness and coherence, and was able to smoothly and deeply discuss open domain topics. Chat, won five championships in the DSTC-9 competition.
Google is also continuing to promote research on basic technologies. For example, in the field of machine learning (ML) algorithms, Google has developed a self-supervised and semi-supervised learning technology called SimCLR, which can simultaneously maximize the consistency between different transformation views of the same image and minimize the transformation view of different images. Consistency between. In the field of machine perception, CvxNet, 3D shape deep implicit functions, neural voxel rendering, CoRNet and other algorithm models have been introduced, and practical applications in outdoor scene segmentation, 3D human shape modeling, image and video compression and other scenes.
In terms of machine learning, Microsoft announced an exclusive cooperation with OpenAI, an artificial intelligence industry promotion agency, to create a top five supercomputer in the world, specifically for training ultra-large-scale artificial intelligence models on the Azure public cloud.
3. Constantly increase the number of cutting-edge technologies such as quantum computing and autonomous driving
The importance and urgency of quantum technological innovation have been widely recognized by all sectors of society. The research progress of quantum computing will also feed back artificial intelligence and cloud computing, adding momentum to the development of new infrastructure. Looking at the world, countries are vying for the first place in the field of quantum computing. The United Kingdom officially launched the "National Quantum Technology Plan" in 2015, and has invested more than 500 million pounds to establish a quantum technology research and development center; in 2016, the European Union released the "Quantum Declaration" flagship project with a total investment of more than 1 billion euros. The United States passed the "National Quantum Action Plan (NQI)" bill in 2018, and plans to increase investment in quantum information science by US$1.275 billion within four years to ensure that the United States stays ahead in the era of quantum technology.
Google is able to defend its dominance in quantum computing, and to pursue the victory. A year ago, Google’s Quantum AI team used quantum computers to surpass the world’s fastest classical computers. In 2020, Google launched a new open source quantum simulator qsim, which will help researchers develop quantum algorithms. The simulator has proven to be able to simulate a 32-qubit quantum circuit with a depth of 14 gates in a Google Cloud node in 111 seconds.
IBM also realized the importance of the quantum field in time and deployed quantum to its traditional advantage cloud computing. Starting to provide quantum secure cryptographic services on its public cloud in 2020. In order to help customers achieve quantum security protection of data in IBM Cloud, IBM will enhance its use of TLS/SSL in its cloud services, aiming to use open standards and open source technologies to achieve quantum security algorithms.
Another milestone breakthrough in domestic quantum computing is Baidu's "Quantity Yifu". In September of this year, Baidu released the country's first cloud-native quantum computing platform—Lianyifu, which realized the deep integration of quantum computing and cloud computing and provided developers with a platform for quantum computing. At the same time, Baidu upgraded its quantum pulse cloud computing service system Quanmai and quantum machine learning development tool set Quandao. Among them, the quantum paddle is a quantum machine learning tool set based on Baidu Feida's research and development, which helps the integration between artificial intelligence and quantum computing.
Baidu Apollo's first 5G cloud driver, the picture comes from the Internet
The development path of intelligent transportation has its own characteristics in various countries, and high-tech companies such as Tesla, Microsoft, Baidu, and Apple also have their own expertise in the development of autonomous driving. With the large-scale deployment of China’s 5G infrastructure and the integration of domestic 5G and AI, technology-based companies such as Baidu’s AI companies continue to build infrastructure such as “vehicle-road synergy”, and more importantly, “all things intelligent—— The world's first fully driverless live broadcast was completed at the "Baidu World 2020" conference.
Technical control "Tesla" will adhere to the technological optimization route in 2020. In October 2020, Tesla released FSD Beta, a fully automated driving test system. FSD is technically the master of the first three generations. Unlike China relying on 5G to transmit data, Musk's aversion to lasers and radars makes "visual algorithms" Tesla's biggest feature. The neural network on its chip can simultaneously process image data generated by 8 cameras. This kind of data algorithm learns and trial-and-error through the analysis of user data, but compared with lasers and radars, this "visual algorithm" also poses potential dangers. For example, in January 2020, Tesla will be "white "Truck" was identified as "cloud", which caused a false acceleration and directly hit the truck.
It is believed that the new round of smart travel technology competition between China and the United States will continue in 2021.
4. Deep integration of cloud and AI
Under the epidemic, the world is accelerating the process of informatization, digitization, and intelligence in all walks of life. As an information infrastructure, cloud computing is on the top of the wave. And we see that the integration of cloud computing and AI is getting deeper and deeper, not only using cloud computing to save digital costs for various industries, but also using AI to help improve the efficiency of enterprises and carry out intelligent transformation and upgrading.
In February, Baidu Smart Cloud, an important carrier and exporter of Baidu AI to B, carried out a strategic upgrade: based on cloud computing, with artificial intelligence as the starting point, focusing on important tracks. In December, Baidu Smart Cloud further proposed the “integration of cloud and intelligence”, which combines technological breakthroughs and applications with the actual needs of the industry, allowing customers to use AI conveniently and quickly like using hydropower and coal, and promote the intelligent transformation of the industry.
Baidu Smart Cloud, integrated with cloud intelligence, has been accelerating the intelligentization of the industry and has achieved many breakthroughs. For example, it cooperates with Beijing Haidian District to build the city brain of Haidian, cooperates with Guiyang Economic Development Zone to build an AI industrial Internet platform, and cooperates with China National Building Materials Group to build smart logistics.
In December 2020, Amazon held the re:Invent2020 conference. Amazon AWS CEO Andy Jassy proposed that in the cloud service market, the following four are disruptive services that can change the rules of the game: database services; data analysis services; machine learning services; and development tools. It can be seen that AI basic technologies such as machine learning are deepening into the Amazon cloud.
5. Under the new infrastructure policy, China's AI development is taking advantage of the trend
On January 20th, Tsinghua University Artificial Intelligence Research Institute and Tsinghua-Chinese Academy of Engineering Knowledge Intelligence Joint Research Center jointly released the "Artificial Intelligence Development Report 2020". The report found that China's scientific research output level in more than 10 AI sub-fields such as natural language processing, chip technology, machine learning, information retrieval and mining is among the highest in the world; the output of papers in the field of multimedia and Internet of Things exceeds that of the United States, Number one in the world; and in the fields of human-computer interaction, knowledge engineering, robotics, computer graphics, and computational theory, China still needs to catch up.
In 2020, after the country put forward the "new infrastructure" policy, new technologies represented by 5G, Internet of Things, cloud computing, artificial intelligence, etc., set off a new round of technology upsurge in the country, and supported the economy to move towards high-quality development. Baidu is the first in the industry to propose new AI infrastructure, relying on new AI infrastructure including Baidu brains, flying paddles, smart clouds, data centers, etc., to promote smart transportation, smart cities, smart finance, smart energy, smart medical, industrial Internet and Intelligent manufacturing and other fields realize industrial intelligent upgrading. I believe that with the joint efforts of the government and enterprises, in 2021, my country's AI development will take advantage of the trend.
In addition, in 2021, AI will continue to integrate with technologies such as life sciences, cloud computing, and the Internet of Things. Chips and quantum computing will become the core of competition among countries. We have also noticed that the development of technology must conform to social ethics, and more laws and regulations will be improved to promote the good use of technology.
(Author: Jiachang Pan; editor: SV Insight)
Reference materials:
https://baijiahao.baidu.com/s?id=1688842290134252551&wfr=spider&for=pc
https://ai.googleblog.com/2021/01/google-research-looking-back-at-2020.html#Quantum
https://deepmind.com/blog/article/alphafold-a-solution-to-a-50-year-old-grand-challenge-in-biology
https://apollo.auto/docment/Autonomous_Driving_Connected_Road_Efficient_Mobility.pdf
https://news.163.com/21/0120/17/G0Q81B8J00019OH3.html
1. In the era of AIoT, new opportunities and challenges of computer technology
2.What is the difference between STM32F1 and GD32F1?
3. Ten tips for placing PCB components, absolutely useful!
4. Intel returns to the technology stream, and former CTO Pat Kissinger returns to serve as CEO!
5. Who owns the RISC-V architecture MCU with embedded professional interface?
6. In-depth analysis of the main function of C language!
Disclaimer: This article is reproduced online, and the copyright belongs to the original author. If you are involved in copyright issues, please contact us, we will confirm the copyright based on the copyright certification materials you provide and pay the author's remuneration or delete the content.