Adhering to the concept of "technology as the basis and innovation as the soul", the Cutting-edge Security Research Center of Topsec carries out innovative research on applications of artificial intelligence (AI) technologies such as machine learning, knowledge graph, and natural language processing and attack and defense technologies such as adversarial machine learning and privacy computing for security issues in the Internet, Big Data, cloud computing, IoT, and industrial Internet fields. The Cutting-edge Security Research Center has profound academic research and engineering transformation capabilities and can provide continuous and stable technical support for the development of Topsec's products and services.
It researches applications such as malicious code detection, threat traffic analysis, and encrypted traffic identification based on machine learning (deep learning) and builds a cybersecurity machine learning model development platform to solve problems encountered by machine learning applications such as low data set quality, low manual labeling efficiency, and model development difficulty.
It researches the natural language processing mechanism in the cybersecurity field by focusing on intelligent word segmentation, entity identification, relationship extraction, text classification, keyword extraction, automatic summarization, and other text processing capabilities in the field of threat intelligence, and builds a threat intelligence processing and analysis platform to solve problems such as low analysis efficiency for a large amount of threat intelligence and difficulties in event tracking.
It researches the knowledge graph construction mechanism in the cybersecurity field by focusing on representation, extraction, fusion, inference, retrieval, and analysis of knowledge, and develops a security knowledge graph application system to solve problems such as automatic discovery of cybersecurity knowledge, balance between security knowledge matching efficiency and extraction accuracy, reliability of security knowledge, and complexity of security knowledge use.
It researches machine learning fragility and adversarial attack and defense mechanisms in adversarial environments by focusing on security attack and defense technologies for machine learning data, models, and algorithms, and builds a cybersecurity-oriented machine learning model evaluation platform to enhance the robustness of cybersecurity machine learning applications and reduce security risks of these machine learning applications.
It researches privacy protection mechanisms in data technology fields such as Big Data and AI by focusing on key technologies such as differential privacy, multi-party security computing, homomorphic encryption, and trusted computing, and builds privacy-protected cybersecurity applications to solve the contradiction among data confidentiality, integrity, and availability and reduce the difficulty in implementing privacy computing applications.
This lab has presided over or undertook the research and development of a series of national, provincial, and ministerial scientific research projects, involving cyberspace security, Big Data security, IoT security, and industrial Internet security and has formed a batch of high-quality research achievements.
