Research Direction

Shenzhen Research Institute of Big Data (SRIBD) will closely focus on the national strategic positioning in the direction of big data and artificial intelligence, serve the major needs of Shenzhen in areas such as big data, artificial intelligence, future new computing, large-scale intelligent system technology and industrialization, work hard to make a significant contribution to big data and artificial intelligence science, technology and industrial applications. The institute will focus on five key research areas: big data foundation theory, artificial intelligence basic theory, future new computing system, big data artificial intelligence application theory and data-driven large-scale intelligent system.

  • Big data foundation theory

    Big data has its own characteristics and development. Studying big data theory can help us to understand big data in essence, which is the key to solve the big data problem in the future. The research of big data involves three aspects: data collection, storage and processing. The main research goals of the institute are (1) to set up a big data collection method that can effectively obtain useful data, further protect the privacy and ensure the correctness of the data; (2) to develop a set of big data storage theory to meet the needs of future unlimited storage and processing of data; (3) to develop a set of new theories, new models, new technologies and software platforms for analyzing and processing unstructured big data.

  • Artificial intelligence basic theory

    Artificial intelligence, accompanied by the advent of computer science, aims to explore the nature of intelligence and produce a new type of intelligent machine that responds in a manner similar to human intelligence. The basic theory of artificial intelligence includes the basic theory of deep learning, the theory of new machine learning mechanism and the foundation of new artificial intelligence system. The main research goals of the institute are (1) to develop a new set of Generative Deep Model (DGM), which enables the model to efficiently learn the essence of the data; (2) to put forward a new type of machine learning mechanism to improve the data adaptability and optimization algorithm, and increase the sensitivity of machine learning to the causal relationship between the data; (3) to establish a theoretical system for the next generation of artificial intelligence.

  • Future new computing system

    The future computing system needs to provide powerful computing power, storage capacity and new computing models to meet the development and application of big data and artificial intelligence. The work of the institute includes three aspects: the calculation models and algorithms for the E-class computing system, the new computing system for big data processing and the new computing system for artificial intelligence. The goals are (1) to build a multilevel, multi-granularity and scalable parallel computing model that adapts to the domestic E-class computer and to study the optimization method for the domestic heterogeneous and multi-core processors; (2) to propose some technologies for large-scale asynchronous distributed parallel analysis and calculation, to efficiently solve the big data computing problems of massive scale, unstructured, heterogeneous sources, security and privacy sensitivity and other difficult characteristics; (3) to build a computing service platform dedicated to big data and artificial intelligence.

  • Big data artificial intelligence application theory

    Two of the most representative applications of artificial intelligence are computer vision and natural language processing. In addition, high-precision prediction with the use of audio, video, picture and spatial and temporal data obtained by sensor networks is the future trend. The institute's work focuses on computer vision, natural language processing and cross-media data fusion. The main research goals are (1) to develop a set of data-driven basic theory of intelligent visual computing to promote the development of robotics, driverless vehicles, interactive, virtual reality, medical diagnosis and other applications; (2) to conduct research on natural language processing, focusing on the syntax of natural language logic, the representation of character and depth of semantic analysis, to promote effective communication between human and machine, and to achieve multi-style multi-lingual and multi-field natural language intelligent understanding and automatic generation; (3) to propose novel multimedia data fusion algorithm, and apply them to some smart applications, including smart campus and data-driven wireless networks in the future, in order to accelerate the industrialization of artificial intelligence technology.

  • Data-driven large-scale intelligent system

    The rapid growth of big data technologies has made it possible to build complex, intelligent systems based on data-driven approaches. Focusing on data collection, processing and system construction, the institute studies large-scale intelligent systems in the areas of smart wireless networks, smart grids, smart finance and smart healthcare. The main research goal is to fully tap the massive data and use data-driven adaptive artificial intelligence technology to integrate data science, artificial intelligence and communication networks, power grids, finance and health, and to design next generation smart system with the ability of independent evolution.