Tag: peking university

Tencent, Huawei, and Baidu Executives to Join China’s New National Blockchain Committee to Set Industrial Standards

The Chinese central government has put together a national blockchain committee to work on setting industrial standards. The Ministry of Industry and Information Technology (MIIT) issued a notice on April 13 of the “Public Notice on the Formation of a National Blockchain and Distributed Ledger Technology Standardization Technical Committee.”

The committee would be made up of 71 individuals from different backgrounds, including political, industrial, academic, and research organizations. The committee will be chaired by MIIT deputy minister Chen Zhaoxiong, along with five vice-chairs, all of whom are government staff, including Di Gang, the vice-head of the Chinese central bank digital currency institute.

Other committee members include executives from well-known Chinese institutions, including Baidu, Tencent, Huawei, Peking University, Tsinghua University, Fudan University, amongst others. The ministry is also asking for public feedback on the committee members until the deadline of May 12, 2020.

China’s take on blockchain

Chinese President Xi Jinping announced in October 2019 that its nation should accelerate the development of blockchain technology. He highlighted that the application of blockchain technology has extended to digital finance, the internet of things, supply chain, and digital asset trading, amongst other areas. The president stressed that it is necessary to strengthen fundamental research of blockchain technology to enhance original innovation ability, for the nation to take a leading position in the blockchain industry.

China has also been researching the application of blockchain and artificial intelligence in cross-border financing, as the deputy head of the State Administration of Foreign Exchange in China announced that there are plans to use these technologies shortly after Facebook announced its plans for its Libra stablecoin. 

Huawei taking on the blockchain road

The Nanshan government of Shenzhen, in the southern Guangdong province in China, has recently announced an agreement with electronics giant Huawei Technologies to establish an information technology center, accelerating the application of technologies including blockchain, artificial intelligence in the district.

With joint efforts between the Nanshan government and Huawei will explore FinTech solutions and development to speed up the application of disruptive technologies. Financial applications, including banks, insurance, securities, and third-party payments, will also be explored based on Huawei’s Kunpeng structure.

Image via Shutterstock

Enhancing AI's Operational Efficiency: Breakthroughs from Microsoft Research and Peking University

In a collaborative effort, researchers from Microsoft Research and Peking University have made significant strides in advancing the capabilities of Large Language Models (LLMs), particularly in the realm of complex instruction following and graphic design generation. This research not only uncovers the limitations LLMs face in operating within complex systems but also proposes innovative solutions that could redefine their application in various fields.

Key Developments and Innovations

WizardLM and Evol-Instruct: The team introduced WizardLM, powered by their novel Evol-Instruct method, which enables LLMs to automatically generate vast amounts of instruction data with varying complexity levels. This approach significantly enhances LLMs’ ability to follow complex instructions, outperforming traditional models and even showing superiority to human-generated instruction datasets in certain aspects​​.

COLE – A Hierarchical Generation Framework: Another groundbreaking project is COLE, developed to address the challenges in graphic design generation. COLE simplifies the process of converting simple intention prompts into high-quality graphic designs by employing a hierarchical generation approach. This involves understanding intentions, arranging and improving visuals, and ensuring quality through comprehensive evaluations. The system demonstrated its capability to produce excellent quality graphic design graphics with minimal user input, marking a notable advancement in autonomous text-to-design systems​​.

Implications and Future Directions

These innovations highlight a significant leap towards enhancing the operational efficiency and versatility of LLMs in performing tasks that require understanding and following complex instructions, as well as generating high-quality graphic designs. By overcoming the limitations associated with manual data generation and the challenges in graphic design, these models pave the way for more autonomous, accurate, and efficient AI applications across various domains.

Exit mobile version