Developer Community Tech Week is back with us again, so check out this week’s important news for us developers.

In one minute!

  • Cctv.com launches grassroots governance big data platform to open up the “last meter” of government services
  • RedMonk programming language rankings: JS continues to dominate, Dart is rising fast!
  • Edge and Brave will match Chromium’s four-week release schedule
  • PyTorch 1.8 released with AMD support, optimized for mass training
  • The world’s first artificial intelligence earthquake monitoring system has been launched with real-time warning feedback improved by one second
  • Red Hat OpenShift 4.7 is coming, further optimizing traditional and cloud native applications
  • AAAI 2021 | Forecasting City Sales Volume using deep meta-learning
  • CVPR 2021 human body image generation and editing method with shape and grain understanding coupling

News of the industry

1. Cctv.com launched a grassroots governance big data platform to open up the “last meter” of government services

Recently, platform of big data of grassroots management of CCTV net goes online formally. This platform is a major measure for mainstream media to fully implement the requirements of the central government to strengthen the governance of litigation sources and build a new pattern of modernization of grassroots governance system and governance capacity. It is of great significance for promoting the goal of “social governance, especially grassroots governance, to be significantly improved” during the 14th Five-Year Plan period. Cctvnet Big Data Research Institute is a digital think tank based on the resource advantages of China Media Group and CCTVnet. Comply with the national strategy of big data layout to the construction of digital China policy as the guide, institute for giving full play to the advantages of CCTV network’s own technology innovation, combined with domestic and foreign digital industry resources, focus on the big data applications, the digital transformation, the new media, new technology applications, network development trend in the field of space governance, With consulting solutions, index research, big data platforms and AI technology products as carriers, it provides decision-making reference and think tank services based on big data for central ministries, commissions and local governments, and contributes to the modernization of national governance system and governance capacity.

2.RedMonk programming language rankings: JS continues to dominate, Dart is rising fast!

RedMonk analyzed the GitHub Archive and Stack Overflow data tools to release the January 2021 (Q1) programming language rankings, which break down the latest changes in the major programming languages.

It mainly includes the following points:

  • The ranking of programming languages has undergone a series of changes recently, and there may be more changes in the future;

  • JavaScript has topped the list for seven consecutive years since 2014, and will continue to lead the list.

  • Ruby is down in the rankings, overtaken by CSS and TypeScript;

  • TypeScript, R, Kotlin, and Rust are moving up the rankings, and Dart is moving up fast and is on track to break into the Top 20.

* For details, please refer to:

Redmonk.com/sogrady/202…

3.Edge and Brave will match Chromium’s four-week release schedule

Earlier this month, Google announced that the Chrome/Chromium release cycle would be reduced from six weeks to four. It will adopt this release cycle starting with the RELEASE of Edge 94 in the third quarter of this year. Chromium-based browsers MS Edge and Brave have both announced that they will match the release schedule. Microsoft says that as a contributor to the Chromium project, it also looks forward to getting new changes to users more quickly through the four-week release cycle. Like Google, Microsoft plans to offer longer release cycles for enterprise customers, with Extended Stable releases coming out with major updates every eight weeks.

4.PyTorch 1.8 released with AMD support, optimized for mass training

Recently, the official PyTorch blog released version 1.8. According to the official release, the new version mainly includes compiler and distributed training updates, as well as some new mobile tutorials. Overall, this update covers more than 3,000 commits since the 1.7 release, including compilations, code optimizations, scientific computing front end apis, and binary support for AMD ROCm via Pytorch.org. PyTorch 1.8 also includes feature improvements and gradient compression for large-scale training of pipeline and model parallelism.

* Tutorials and documentation details can be found at:

Pytorch.org/blog/pytorc…

5. The world’s first artificial intelligence earthquake monitoring system has been launched with real-time warning feedback improved by 1 second

A team from the University of Science and Technology of China, in cooperation with the China Earthquake Administration, has launched the world’s first artificial intelligence earthquake monitoring system, the “intelligent ground motion” monitoring system, which can accurately estimate earthquake focal mechanism parameters within one second, China New Video reported. It is reported that the occurrence of earthquake is the rupture and dislocation process of rocks at the focal point, and it is a time-consuming process to calculate the focal mechanism of earthquake from the seismic records. Based on millions of seismic data collected in the database and combined with seismological theory, the system can quickly process the ongoing seismic data. It can not only infer the rupture direction and velocity of faults, but also predict the possible distribution of tsunamis and strong aftershocks.

The results, recently published in Nature Communications, will soon be tested on the “Intelligent Ground Motion” earthquake monitoring system jointly developed by the University of Science and Technology of China and the China Earthquake Administration.

6. Red Hat OpenShift 4.7 is coming to further optimize traditional and cloud native applications

Red Hat, the world’s leading open source solution provider, has announced the release of the latest version of its industry-leading enterprise Kubernetes platform — Red Hat OpenShift 4.7. Based on Kubernetes 1.20, this release is designed to simplify and accelerate application modernization and help IT teams eliminate barriers to unified management of traditional and cloud-native applications. All new features will be implemented on the Red Hat OpenShift consistent platform in an open hybrid cloud environment. Red Hat OpenShift offers a variety of application modernization and migration capabilities to meet the specific needs of enterprises. The latest release adds more modern tools for migrating legacy applications to a new microservices-based architecture, enabling unified management of all applications to improve operational efficiency without compromising innovation.

The academic frontier

1.AAAI 2021 | Deep element learning is used to forecast urban sales volume

In this paper, the author also made a comparison of the time-space alternation training method in three large shopping festivals. In time training, merge operations integrate instances from the same day type in batches, regardless of spatial knowledge. Meanwhile, in spatial training, the model integrates a batch of instances from the same area without considering the time information. However, the lack of joint meta-knowledge in space and time still leads to lower accuracy compared with complete alternate training. This suggests that different training patterns learn different transformations across regions and time types. The temporal and spatial supplementary training can effectively improve the accuracy of prediction.

* Paper links:

Urban-computing.com/pdf/AAAI202…

2.CVPR 202 | Human body image generation and editing method with shape and grain understanding coupling

In recent years, it has been a hot topic in the field of vision to generate images from a single human body image in any Angle of view and any posture. The existing methods can not achieve flexible image editing and it is difficult to reasonably predict the invisible information. The fundamental reasons are the coupling of clothing shape and style/texture and the loss of spatial information. In order to solve the above problems, the research designed the shape and style/texture information separation scheme, established the stage area image representation model, joint global and local information reasonable prediction of target image, at the same time use the spatial perception of regularization method to retain information, realize the new posture of the semantic boot image generation and editing, It breaks through the bottleneck of shape and texture decoupling, and endows the algorithm with flexible and controllable editing ability.

* Links to papers: arxiv.org/abs/2103.04…

The above information is from the network, edited by the “JINGdong Technology Developer” public account, does not represent the position of Jingdong Technology

Recommended reading

  • Jingdong cloud new generation from the cloud server online in April; COLING 2020 | Bidirectional cognitive thinking Network for machine reading Comprehension
  • Blockbuster! Jingdong cloud since the fourth generation cloud host release; Foreign physicists have developed assembly language for quantum computers
  • Northwestern University developed monkey face recognition technology. Golden monkeys can swipe their faces to punch in. IJCAI 2020 | Text style Conversion based on hidden space of learning instance

Welcome to [JINGdong Technology] to learn about the developer community

More wonderful technical practice and exclusive dry goods analysis

Welcome to “JINGdong Technology Developer” public account