What Has Alibaba’s DAMO Academy Achieved In Three Years?


Alibaba DAMO Academy (Academy for Discovery, Adventure, Momentum and Outlook) just welcomed its third anniversary.

As Alibaba’s technology research institute for basic scientific research and core technology development, it claims to have invested hundreds of billions of yuan in three years to develop core basic technology.

What has it achieved in the past three years? Did it meet its goals?

The purpose of the DAMO Academy is use technological innovation to promote industrial innovation, cultivate scientific and industry talents, and explore the establishment of "research-to-engineering-to-industry " innovation system.

It aims to create an architecture combining cutting-edge technologies (such as quantum technology), breakthroughs in core technologies (such as AI, chips), and the application of key technologies (such as databases), to formulate different levels of goals, while frequently assess progress and adjust incentive mechanisms.

DAMO Academy has established the "4+X" research framework, namely machine intelligence, data computing, robotics, financial technology, combined with "X" research directions.

In 2018 and 2020, it has incubated Pingtou Semiconductor, a chip company, and Xiaomanlu, a logistics robot company. It operates 14 laboratories in total today.

Its research centers have been established in 8 regions around the world, including Hangzhou, Beijing, Shanghai, Singapore, Israel, Seattle, and Silicon Valley.

The average age of full-time researchers is 32 years old, and the post-90s generation accounts for 34% of the total. There are more than 30 professors from well-known universities and more than 10 IEEE Fellow-level scientists employed at DAMO Academy.

A List Of Achievements

The three-year results of DAMO Academy can be examined from these three dimensions: Basic scientific research, industrial applications, and improving livelihood.

1. Basic scientific research

DAMO Academy has won more than 60 world firsts in top international technical competitions and published more than 100 international top conference papers.

A number of social activities on basic scientific research have been initiated, especially competitions and awards for young talents.

For example, the public welfare event "Alibaba Global Mathematics Competition" for math enthusiasts around the world has attracted nearly 100,000 participants from around the world.

Another example is the establishment of an award for young Chinese scientists aged 35 and above: the Green Orange Award. It has been held for three times, and a total of 29 young scientists have received 1 million yuan prizes.

In addition, it has made breakthroughs in core basic technology.

Such as the ultra-large-scale graph computing system Graphscope, the quantum circuit simulator "Tai Zhang 2.0″, the new storage-computing integrated architecture.

Take its quantum laboratory as an example. "Tai Zhang 2.0″ is the first large-scale quantum development platform (ACQDP) open sourced by the DAMO Academy Quantum Laboratory.

It uses the distributed tensor network shrinking algorithm library as the calculation engine, which can realize large-scale quantum circuit simulation and help the research and development of large-scale quantum computers and quantum hardware.

At the beginning of 2020, Taizhang 2.0 also simulated the quantum circuit that Google claimed to achieve "quantum superiority". The entire task was compressed to 20 days to complete, an improvement of 4 orders of magnitude over the previous plan.

Currently, the DAMO Academy team has further launched the construction of superconducting quantum hardware projects.

Shi Yaoyun, the head of the Quantum Lab, revealed that the next breakthrough lies in the quantum hardware project.

2. industrial applications.

The most obvious industrial incubation of the DAMO Academy in the past three years has been the establishment of Pingtouge Semiconductor and the Xiaomanlu logistics robot.

Pingtouge is the first company officially incubated by DAMO Academy. In September 2018, Alibaba announced the integration of Zhongtian Micro and DAMO Academy to establish a chip company: Pingtouge Semiconductor.

In less than a year, its first chip was successfully delivered. Xuan Tie 910 and Hanguang 800 are the two chips coming out of Pingtouge.

In addition, DAMO’s internal visual laboratory has realized 4 business applications of its technology.

According to Xu Yinghui, the person in charge of the Vision Lab, the lab currently uses its technology on digital intelligence e-commerce, digital intelligence media, digital intelligence earth, reading light OCR (text image understanding).

In Digital Earth, the remote sensing AI platform AI EARTH is one of the platforms out of the Vision Lab.

AI Earth uses deep learning, computer vision and other technologies to integrate and analyze multi-source earth observation data such as satellite images, drone images, real-time video streams, meteorological data, IoT data, to achieve digitization of multiple types of buildings, land, vegetation, and rivers.

If someone wants to build a physical extraction of a single prefecture-level city, the traditional manual method may take several months. But AI Earth takes a few minutes to complete the task, and the target detection accuracy rate can reach more than 95%.

During the national flood season in July last year, the Vision Lab developed and launched the SAR radar image water recognition algorithm in only 7 days, and the number of images processed was 5 times higher than usual.

Traditional weather forecasting methods rely on large-scale meteorological physics calculations, but this method cannot make short-term forecasts.

What Xu Yinghui’s team does is to make weather forecasts for 2-3 hours, as the uncertainty level will not be so high.

3. improving livelihood.

DAMO Academy has the ambition to solve "social challenges". The most representative results is its medical AI tool.

At the beginning of this year, when the pandemic broke out, DAMO Academy became the label of Alibaba AI’s pandemic fighting efforts.

It conducted genetic testing platform cooperating with Zhejiang Center for Disease Control and Prevention, to reduce the time for genetic analysis of suspected cases to half an hour.

It is also the first CT imaging system to be applied in various places, and it has been deployed in more than 600 hospitals in Zhejiang, Henan, Hubei, Shanghai, Guangdong, Jiangsu and other cities, as well as in Japan, Indonesia and other countries.

Up to now, the CT imaging system has diagnosed a total of 800,000 clinical cases.

Its mass produced robot, Xiaomanlu, is low-cost and reliable. It can run more than 100 kilometers with only 4 kilowatt-hours of electricity. It can deliver up to 500 express delivery per day. It can handle extreme environments such as thunderstorms, lightning, high temperature, rain and snow, and navigate garages and tunnels.

Can The DAMO Academy Model Be Copied?

The research model exhibited by DAMO Academy is the parallel development of basic research and industrial applications.

Therefore, DAMO Academy scientists are no longer limited to solving pure academic scientific research problems. They also need to think about how to meet the needs of engineering.

For example, DAMO Academy scientists need to go to the business frontlines at Alibaba and understand the needs of the industry and users. Their performance is not measured by just publishing research papers.

The DAMO Academy model is based on its orientation as demand and problem-driven research. Its research starts with a real world problem.

Then, it focuses on scientific research with a mind of putting the research for solving a business problem and finding a commercialization angle.

Overall, it has dual objectives of conducting research in support of industrial applications. It balances and coordinates the two objectives to achieve both scientific advances while generating business values.

This kind of corporate research lab model has so far prove to be more practical and brings more value to the parent corporation compared to other technology companies.

 
Introducing Finley, the fintech newsbot from China Money Network! Using state-of-the-art machine learning technologies, Finley scours the globe for immediately usable data about the latest private equity deals, venture capital funding rounds, and business expansions happening in China. Finley asks: Who's buying whom? What company just raised new capital? What is the relationship among the parties? Then Finley uses his very real artificial intelligence to decipher that unstructured data to write news articles for you to read. So from unstructured data to a news story you can understand, China Money Network's Finley is your fintech newsbot business partner!

RELATED NEWS



LEAVE A REPLY