Microsoft’s newly established AI4Science team, machine learning giant Christopher Bishop is the director, and Liu Tieyan is the head of Beijing

90
Frontier of technology

Microsoft Research established a new scientific intelligence team AI4Science: Christopher Bishop as director, Liu Tieyan as head of Beijing

Recently, Kevin Scott, executive vice president and chief technology officer of Microsoft, announced the establishment of a new scientific intelligence team AI4Science, which brings together many world-class experts in machine learning, computational physics, computational chemistry, molecular biology, software engineering and other disciplines. Committed to making the Fifth Paradigm a reality.

Among them, Christopher Bishop, dean of Microsoft Cambridge Research Institute, is the director. Christopher Bishop is an honorary professor of computer science at the University of Edinburgh, a researcher at Darwin College, University of Cambridge, and a member of the Royal Academy of Engineering in 2004. Focus on machine learning and natural sciences.

Liu Tieyan, deputy dean of Microsoft Research Asia, is the head of Beijing. Liu Tieyan graduated from the Department of Electronics, Tsinghua University, and will be an ACM Fellow in 2021. It is reported that the Graformer general molecular modeling model established by Liu Tieyan recently won the championship in the Open Catalyst Challenge. The model has strong molecular characterization capabilities, and will be very useful in new material design and drug discovery. Great help.

In addition to the Chinese and British teams, Microsoft has also set up a new AI4Science laboratory in Amsterdam, and has hired machine learning expert Max Welling to participate in the experiment. Currently, the laboratory is under construction in the Amsterdam Science Park. (Microsoft)

DeepMind AI releases PLATO deep learning system: it can be used to study human learning “intuitive physics”

On July 11, DeepMind AI released a deep learning system called PLATO (Physics Learning through Auto-encoding and Tracking Objects: Physics Learning through Auto-encoding and Tracking Objects), which was jointly developed by computer scientist Luis Piloto and his team. The system it contains is inspired by how infants learn, using animated videos of simple objects like cubes and balls to train neural networks to produce “surprise” behavior in the face of predicted differences in “impossible” events, which can be used to study human learning” Intuitive Physics”.

According to the calculation screen, after receiving about 30 hours of video training, including balls rolling down slopes and balls bouncing between balls, the system faced “impossible” scenarios given by researchers (for example, objects passing through each other did not occur). interaction), PLATO can show intuitive physical reactions such as “surprise” and “surprise”.

According to the research team, PLATO can be used as a powerful tool to study how humans learn intuitive physics, and also shows that object representation plays an important role in human understanding of the surrounding world.

Webb Space Telescope’s first full-color image released: always using Ka-band frequencies

On July 11, NASA announced the first full-color photo (below) taken by the James Webb Space Telescope (JWST). . In what NASA calls the deepest infrared image of the universe to date, the photo shows that galaxy cluster SMACS 0723 contains thousands of galaxies and is 4.6 billion light-years from Earth.

90

But compared to the much-anticipated first full-color photo, the spacecraft communications subsystem used by JWST isn’t flashy. According to JWST staff, the plan has been to use Ka-band frequencies to transmit data back to Earth on a 25.9-gigahertz channel at 28 megabits per second.

Additionally, while the 25.9-GHz Ka-band frequency is the primary communications channel for the JWST, it uses two channels in the S-band: one is the 2.09-GHz uplink, which enables future transmissions and scientific observation programs to be 16 kilobits to the telescope; the other is a 2.27 GHz, 40 kb/s downlink through which the telescope transmits engineering data—including its operational status, system health, and other information about the telescope’s day-to-day activities information.

Apple may push “lockdown mode” in September to protect product users from spyware attacks

Recently, Apple announced that it will introduce a “lock” option for iPhone, iPad and Mac computers to protect users of the product from spyware launched by state-sponsored hackers.

The new functionality of Lockdown Mode will be available as a beta release so that security researchers can identify any bugs or vulnerabilities. Of these, Apple typically releases major updates to its device operating systems in late September.

It is reported that “lock mode” is intended to be used as a panic button, and activating the lock will disable many functions including sending attachments and links as text, receiving FaceTime calls from new numbers, and web browsing. Apple expects that only a small number of users will need the button.

Apple believes the extra protections are important for activists, journalists and other targets of hacking by well-funded groups. Users will be able to activate and deactivate Lockdown Mode at will. (Associated Press)

Landing pioneer

Zhujian Intelligence received D1 round of financing of nearly 100 million yuan

Recently, Zhujian Intelligence announced the completion of the D1 round of financing of nearly 100 million yuan, which was jointly invested by Jinku Capital, Jiangsu Cultural Investment, and Junci Capital. The financing funds will be used for the promotion of SaaS standardized products, continue to consolidate technical barriers and product advantages, and help more Multi-head and waist customers realize the intelligent transformation of enterprises.

According to public information, Zhujian Intelligence is an NLP and multi-modal emotion recognition technology developer founded in 2015 by Jian Renxian, the former vice president of Microsoft (Asia) Internet Engineering Institute. Jian Renxian, former vice president of Microsoft (Asia) Internet Engineering Institute, was responsible for Windows Mobile, Bing search, advertising, and global mobile products R&D business, and his total Google Scholar citations reached 68,000.

Since its establishment in 2015, Zhujian Intelligence has covered AI+finance, AI+enterprise, AI+healthcare, AI+manufacturing, AI+intelligent terminal, AI+government and other fields. →Code-free autonomous operation and maintenance” systematically cultivates smarter digital employees, and has launched three core products, Emoti X, Emoti V in the interactive AI platform, and Gemini, a cognitive knowledge platform.

Previously, Zhujian Smart had completed C+ rounds, C rounds, and B rounds of financing in 2021, 2020, and 2019.

Heisenberg Robotics Receives Multi-Million Dollar Angel Round of Financing

AI smart courtyard solution provider “Heisenberg Robotics” received a multi-million dollar angel round of financing, led by DCM, followed by Jiangmen Ventures, and Wanshi Capital acted as the exclusive financial advisor for this round of financing; the financing funds will be used In order to continue to increase the research and development of core technologies such as AI vision algorithms, a new generation of modular intelligent garden robots has been launched.

According to public information, Heisenberg Robot was established in early 2022. The team’s genes are autonomous driving. The founder and CEO Wu Xingyun worked for L4 autonomous driving company “Uisee Technology” and was deeply involved in the research and development of visual theoretical algorithms and engineering products. , In addition, many core members are from companies such as Uisee, Xiaomi, ZTE, and Huawei, and R&D personnel account for 90%.

“Heisenberg Robot” adopts a full-vision technology route. The product focuses on European and American courtyard scenes. In addition to providing lawn mowing functions, it also provides users with functions such as leaf blowing, edge mowing, cleaning and sprinkler irrigation through detachable modules.

Leifeng Network

This article is reprinted from: https://www.leiphone.com/category/academic/xjf9r1nKDXS7PWrI.html
This site is for inclusion only, and the copyright belongs to the original author.

Leave a Comment