AI Inclusion and Security System Research Center

The AI Inclusion and Security System Research Center is based on distributed computing, blockchain, cryptography and edge intelligence and focuses on secure, fair and inclusive platforms for data use and regulation, platforms for secure sharing of personal data and privacy assurance systems.

Dr. Jiaping Wang serves as Chair Scientist in the AI Inclusion and Security System Research Center. His previous positions include Research Director in Microsoft Research (headquarters) and Technical Partner in Sinovation Ventures and Deputy Director of the AI Engineering Institute. Dr. Wang Jiaping has made dozens of research findings in distributed systems, computer graphics and vision, and GPU clusters of machine learning. Those findings have been published on ACM/ToG top-level journals and granted with patents in the U.S..

The vision of the AI Inclusion and Security System Research Center include: to allow the whole society to benefit from the next generation AI technology, in particular to provide technological guarantee for SMEs and individual consumers; to improve the whole society’s abilities for information processing and digital decision-making through secure, fair and inclusive use of intelligent technologies; to enhance individuals’ ability to protect information security so as to avoid social problems caused by malicious large-scale intelligent technologies, such as data abuse, privacy violation, information cocoons and manipulation of group consciousness; to curb the development of new monopoly tools by means of intelligent technologies and their harm to industries and innovation; to provide underlying technical support and innovative technology solutions for regulation technology (RegTech).

Research Findings

The AI Inclusion and Security System Research Center has independently developed the “Moqun” blockchain platform equipped with fully independent intellectual property rights. Compared with traditional blockchain systems, Moqun has delivered the “Asynchronous Consensus Zone” model, extending the traditional single-chain blockchain model to the multi-chain model. Furthermore, it introduces the “ultimate atomicity” technology to enable the effective processing and execution of transactions spanning multiple consensus zones, with the cost irrelevant to the number of shards in the whole network. This model vastly increases the blockchain throughput, the whole-network transaction processing capacity and the memory space of status expression by magnitudes (10^3 times) on the basis of “decentralization” and “security”. In addition, the model ensures neutrality of the upper transaction structure. In other words, it perfectly meets the requirements for large-scale payment and clearing systems in any scenarios without introducing assumptions on any transaction structure.

Under this completely new structure, the AI Inclusion and Security System Research Center has further independently developed the new generation of contract language PREDA (Parallel Relay Execution Distributed Architecture), which substantially increases the efficiency of development and execution and makes it easier to guarantee the security of intelligent contracts. Equipped with fully independent intellectual property rights, PREDA supports the development of user-defined contracts and has a Turing-complete transaction logic. The PREDA language and its compiler have thoroughly stepped out of the current status in which domestic blockchain systems are relying on the programming models and languages of foreign open-source projects. On the basis of fundamental capacity, PREDA supports large integers of up to 8,192 digits and definable, original and divisible digital currencies that do not allow double-spending, namely, “first-class citizen”assets. The execution engine also provides support for periodic transactions, programmable multiple signatures and on-chain random numbers. Meanwhile, it prohibits implicit type conversions at the grammatical level, with higher security for the built-in secure numerical calculation. In terms of performance, PREDA makes full use of the performance of underlying “Moqun” blockchain parallel systems and seals the underlying systems, so that developers can harness high-performance parallel computing without having to worry about the details of underlying parallel computing systems.

To support more flexible business scenarios, the research center’s PREDA fully supports diverse underlying system structures. In other words, the PREDA virtual machine can run on different system platforms and fully support the models “single machine”, “cluster of data center” and “decentralization”.

Research Directions

The high-performance blockchain platform, with blockchain sharding technology as its core, develops large-volume parallel blockchain systems and supports high-throughput, large-volume blockchain platforms.
The research center designs programming languages (DSL) of special fields for parallel blockchain systems as well as the related compiler systems and execution environments to cut the costs of blockchain application and development and reduce program errors in blockchain application. It isolates technical detail of blockchain systems and parallel systems for R&D personnel of traditional information systems to reduce R&D thresholds.
The research center designs universal protocols and underlying platforms for point-to-point distributed communication, storage and computing, providing technical foundation without reliance on internet giants for the application of personal communication and enterprise collaboration.
The research center designs and develops computing platforms for the next generation of trusted computing hardware, providing technical foundation for privacy protection-based large-scale data integration, sharing and computing.