皇冠体育app

Novel Programming and Software Methodologies

With the development of the Internet, cloud computing, big data, artificial intelligence, and the Internet of Things, the ternary integration of human-machine-things and the application scene is the computer have become new trends in the development of software technology. Open, dynamic, and changing requirements and environments pose fundamental challenges to software methodology and technical systems. For this reason, with the support of the National Natural Science Foundation of China's major research program integration projects and innovative group projects, national key research and development programs, nuclear high-tech, and other major projects, the research team of this laboratory proposed to build a pair of New software methods and technology systems that play a leading role in the software field and have achieved remarkable results.

In terms of software paradigm and methodological framework, a meta-level and defined network-structured software paradigm mechanism is proposed and developed to support a new form of software that is perceived, definable, learnable, and adaptable. In terms of open environment perception, the distributed environment perception and efficient environment context detection technology are systematically studied, which enhances the environment perception ability of the software. In terms of system adaptation and evolution, a multi-level, high-consistency, and low-interference software online update technology is proposed, which enhances the seamless evolution capability of the software. In terms of quality assurance of new software, important progress has been made in the analysis and verification of complex program behaviors such as concurrency and distribution, as well as in difficult problems such as concurrent memory access dependencies and the progress of concurrent objects. Based on the above-mentioned theories, methods, and technical achievements, a series of software platforms and tools have been developed, which provide strong support for the development, operation, and adaptive growth of networked software systems.

The above-mentioned original and systematic achievements have produced extensive academic influence at home and abroad. Relevant papers have been published in first-class international journals and conferences, and have won outstanding paper awards in important international conferences such as ASE, ICSE, and PLDI. Internationally renowned journals and publishing houses have published academic works, special issues, and special issues on Internet-based software; the CCF conference has organized a special forum on Internet-based software. Through cooperation with well-known software companies such as Zhongchuang Software, iQiyi, Huawei, and ZTE, relevant achievements have been successfully applied in key fields such as cloud computing, the Internet of Things, and electric power informatization, with obvious economic and social benefits. He has successively won the second prize of the National Science and Technology Progress Award, the first prize of the Ministry of Education's Technology Invention Award, and the Ho Leung Ho Lee Science and Technology Progress Award. The Network Architecture Software Theory, Method and Technology jointly declared with Peking University was selected as one of the Top Ten Scientific and Technological Progresses in Chinese Universities.

Machine Learning and Intelligent Systems

Intelligence is the mainstream trend in the development of information science and technology, and machine learning is the key to realizing intelligence. The fundamental problem in the field of machine learning is how to learn a strong generalization performance model based on existing data information. This goal has been basically achieved under the condition of strong supervision information, but the practical tasks of the next generation of artificial intelligence often face weak supervision information, and machine learning technology urgently needs to break through the serious constraints caused by it.

With the support of National Key R&D Program Project, 973 Program Project, National Natural Science Foundation of China Key Project and Innovation Group Project, National Defense Innovation Special Zone Project, etc., the research team of this laboratory has made a breakthrough in the theory and method of machine learning based on weak supervision information , mainly including: (1) Established a machine learning theory and method based on insufficient information, through the integration of multiple learners to deeply explore the data distribution information to alleviate the lack of sampling information, and lay the foundation for breaking through the bottleneck of insufficient data information; (2) Solved the problem of interval theory that has been debated internationally for fifteen years, and gave a theoretical explanation for the singular phenomenon that AdaBoost does not overfit when supervision information is insufficient, giving birth to a large class of new machine learning methods, the optimal interval distribution learning machine. (3) Created a new machine learning framework for ambiguous objects, opened up a new direction of multi-instance multi-label learning, and broke a new path for learning oriented to ambiguous objects. (4) The first non-neural network deep learning model is proposed, which does not rely on gradient calculation and BP algorithm, and provides a new solution for deep learning without huge training samples.

This achievement has published several high-level papers with important influence, cited by more than 60 countries/regions, including famous institutions such as MIT, Stanford, and CMU, and there are many authoritative scholars such as Turing Award winners, which has triggered a large number of international peers. Follow the research. Some achievements won the second prize of the National Natural Science Award and the first prize of the Natural Science Award of the Ministry of Education. Relevant technologies have been transformed and implemented through the Huawei-NTU LAMDA Artificial Intelligence Joint Laboratory, JD Nanjing Research Institute and Baidu, Ali, Tencent, Didi, etc., providing important support for the breakthrough of the core technology of artificial intelligence in Chinese enterprises, and also providing support for the North-South It is evaluated as excellent for its contribution to major engineering services such as polar scientific research and major national defense needs. The development process of the results has enabled our laboratory to establish a domestically leading research team with important international influence in the direction of machine learning.

Software Trust Assurance and Automation Technology

With the vigorous development and popularization of the Internet, cloud computing, big data, Internet of Things, artificial intelligence, and other technologies, the operating environment of software systems is becoming more and more complex. The open, dynamic, and uncertain characteristics of complex environments and the growing scale and complexity of software systems make software quality assurance more difficult. It is urgent to promote the research on trusted software methods and technical systems in complex environments to a new level.

With the support of major research programs of the National Natural Science Foundation of China, the National Key R&D Program, and the 863 Program, the research team of this laboratory is oriented to the actual application needs of the industry and aims at large-scale complex system research based on software analysis, testing, and verification. Trusted software methods, technologies, and tools have broken through the bottlenecks of existing technologies in terms of bounded reachability verification of hybrid systems, program analysis for dynamic languages, intelligent software testing, prediction and repair of code defects, and effectively Controlling the complexity caused by the scale of the program and the dynamic and uncertain operating environment, a trusted software method and technical system based on cognition and understanding in a complex environment have been formed, which provides an effective means for reducing costs and improving efficiency for software quality assurance in the industry and pathways.

The innovative work of this achievement was published in important academic journals and conferences at home and abroad, such as Science China, ACM TOSEM, IEEE TSE, ICSE, FSE, OOPSLA, ISSTA, RTSS, etc. Best Paper Award and FSE2016 Distinguished Artifact Award; related tools developed have been included in international authoritative textbooks, and have won first place in international tool competitions in related fields, and have been downloaded and used by researchers from the United States, Canada, Germany, etc. (including those from UC Scholars from Berkeley, CMU, UBC, etc., with more than 500 downloads); some of their work won first prizes for scientific and technological progress in Jiangsu Province and Hubei Province; Invention patent transfer/licensing fees exceeded 10 million; the developed online software testing education platform (mooctest.com) attracted more than 20 software companies such as Ali, Baidu, Huawei, etc. More than 16,000 teachers and students of the university provide production-education integration services.

Distributed Ubiquitous Computing and System Security Technology

With the continuous increase in the demand for the ternary integration of humans, machine, and object, the computing platform has changed from functional demand-driven to application scenario-driven. The application bottleneck of storage makes the key supporting technology of integrated distributed computing a major application requirement for the development of the information industry.

Funded by the National 973 Program, the National Key R&D Program, and the National Natural Science Foundation of China, the research team of this laboratory focuses on core areas such as deep data perception fusion, efficient network transmission guarantee, dynamic collaborative scheduling of resources, and reliable, safe, and fault-tolerant storage. Carry out research work on technology and industrial application; focus on multi-dimensional goals such as user quality, system performance, energy consumption, and security, use human-machine fusion perception and privacy protection-ubiquitous network transmission optimization and security assurance-data center resource coordination and reliability assurance As the main line, it realizes the conversion of end device perception from simple single-modality to rich multi-modality, network transmission from simple store-forward to store-process-forward, and data center from focusing on The transformation from resource aggregation and reuse to resource collaborative optimization solve bottleneck problems such as perception fuzzy, transmission limitation, computing imbalance, storage redundancy, etc. Collaborative distributed application support technology system.

The innovative work of this achievement was published in important international journals JSAC, TMC, TIT, TIFS, etc., important international conferences such as SIGCOMM, MobiCom, CCS, INFOCOM, UbiComp, SIGMETRICS, etc., and won the frontier conferences MobiQuitous 2013, ICNP 2015, APNet 2018, etc. The best paper award was selected as a focus paper by IEEE TPDS, which has a wide international influence. Based on the above key technologies, the corresponding distributed support platforms and systems have been developed, and they have been popularized and applied in industries such as electric power, water conservancy, rail transit, and digital media, and have achieved significant economic and social benefits. He has successively won the first prize of the Natural Science Award of the Ministry of Education in 2015, and the first prize of the Science and Technology Award of Jiangsu Province in 2016 and 2019.

Big data media computing and content processing technology

The fusion of human-machine-things is a typical form of the future development of computer application systems. Multimedia information is the basic carrier form of human-computer-things interaction. Domain-oriented and semantic media content processing is the human-oriented and natural way to improve the modern application system of human-machine-things interaction. The important technical development direction and key core support of the degree of globalization.

With the support of 863 major projects, national key research and development projects, national major science and technology special projects, and other projects, the laboratory research team closely combines the characteristics of the application field to deeply explore the methods and technologies of media content analysis, semantic representation, and reasoning applications, through the research on semantic analysis technology of text, image video, and 3D model, gradually build a complete processing flow from perception, cognition to reasoning application, and improve and improve it in typical scenarios, in the domain and semantic media Important achievements have been made in content processing technology and application.

The innovative work of this achievement has been published in important domestic and foreign academic journals such as T-PAMI, TVCG, TIP, Journal of Software, Journal of Computer Science, and international important academic journals such as CVPR, ICCV, AMMMM, WWW, ACL, IJCAI, AAAI, etc. At the meeting, it formed its research characteristics in related technical fields at home and abroad, and carried out extensive cooperative research and promotion and application of results with Huawei, Tencent, Sinopec, and other domestic industry leading enterprises. Among them, the falcons semantic search engine developed by the team is the only one of the three semantic Web search engines recommended by the Internet Consortium (W3C) in China; the key technologies and systems of open domain knowledge association, reasoning, and retrieval research have been used in the real exam questions of the college entrance examination in recent years The experiments on the computer and simulation questions have achieved remarkable results; the computer automatic drawing reading technology has achieved important breakthroughs and remarkable results in the industrial application of architectural drawing recognition, and has achieved good results in several international document recognition competitions. The research results of video representation modeling and object-relationship modeling and analysis technology have won the first and second place in the evaluation of related international fields many times; the results of 3D modeling and intelligent analysis technology for massive seismic data have made great achievements in Sinopec's comprehensive oil and gas interpretation system. Good application effect.

Theoretical Computer Science