With big data technology revolutionizing modern lifestyles through its valuable use in the digital economy, medical fields and e-education to name a few, the Data Science research department of ICT, CAS is dedicated to utilizing big data analysis for better human decision-making and social welfare. Since data-driven intelligence algorithms and deep learning have had a transformative impact on AI tasks, scientific breakthroughs and industrial management, improvements in data technology are called for to ensure algorithm reliability. The Data Science research department’s focus on advancing big data theory, algorithms and systems is directed toward innovation to provide trustworthy intelligence algorithms. The Data Science research department has three affiliated groups, namely Key Laboratory of Network Data Science and Technology, Data Intelligence System Research Center, and Domain-Oriented Intelligent System Research Center.
Overall research themes in our department are web data science, social computing, and swarm intelligence.
Web Data Science
The multidisciplinary field of web data science combines the expertise of computer science, mathematics, and statistics to study complex web data and extract valuable insights. The intent is to transform large, heterogenous data into usable information from which knowledge can be extracted to form wise decisions. The technologies and techniques in this field are geared toward intelligent forecasting, informed decision-making, and support for downstream tasks in various domains. Our research topics span the entire process from raw data to wisdom, including data representation, information retrieval, data mining, open knowledge computing, interactive analysis and causal machine learning. We focus on developing effective models but also on understanding the underlying theory of the tasks, such as causal relations, to ensure models are robust and trustworthy.
The interdisciplinary research field called social computing embraces and draws into its sphere the social sciences, computer science, psychology and physics, among others, for the study of social phenomena. Through computing theory and methods, social activities, human behaviors and interpersonal interactions can be better understood or quantified according to different purposes. Social computing mainly will transition from descriptive research (what happened) to predictive research (what is likely to happen), and then to decision-making research (what should we do). Our research topics cover different development stages of social computing, and our focus is on the challenges of multi-modal, cross-domain, and inter-space considerations. Specifically, we carry out network representation learning and natural language processing in terms of descriptive research; network structure prediction and spatial-temporal prediction in terms of predictive research; and cognitive computing and algorithm auditing in terms of decision-making research.
Evolution in data science, artificial intelligence and wireless mobile network technology has made possible unmanned swarms consisting of aerial vehicles, ground vehicles, ships and other unmanned systems. Their automatic capabilities mean that intelligent unmanned swarms are capable of complex tasks collaboratively across various fields. Improved technologies and techniques make possible the creating of intelligent unmanned swarm systems for different applications, from single member to the whole swarm performance. Topics concerning single agent include embedded AI computing system design, algorithm-hardware co-design for AI, optimization of AI algorithms, remote sensing data processing and analysis. Topics concerning swarm include hard real-time wireless communication, collaborative perception/prediction/decision algorithms, spatial-temporal data sci-tech, and confrontation algorithm for swarms. We are not only committed to solving basic scientific problems and core technical problems, but also to building intelligent swarm systems with next-generation capabilities.