Taiwan Academic Cybersecurity Center

This project mainly explores issues related to information security in fintech. Financial information security issues include three major aspects: AI + Security, zero trust, and post-quantum cryptography and its applications in fintech. In these three major aspects, we propose seven sub-projects to solve different issues and difficulties.

Sub-project 1: Research on homomorphic encryption and zero-knowledge proofs in post-quantum cryptography, as well as their applications in zero-trust domains. We will combine the results with sub-project 5, to elevate authentication technology to post-quantum levels.

Sub-project 2: A privacy-preserving federated recommendation system is an advanced application of artificial intelligence in precision marketing in financial technology. This study aims to develop poisoning attack and defense techniques based on contrastive and adversarial learning to effectively enhance its information security.

Sub-project 3: Focus on two aspects of automated model testing and efficient adversarial example detection. This study aims to contribute to 1) the development of advanced program analysis tools to conduct concolic testing on RNN/LSTM/Transformer models, and 2) the development of advanced XAI-based adversarial example detection tools to protect vulnerable applications from adversarial attacks.

Sub-project 4: We will improve the AI analysis capability by multimodal fusion techniques with heterogenous cybersecurity data, and we adopt the language model to automatically discover the attack patterns for explainable forensics.

Sub-project 5: Discuss the traditional EMV payment architecture to address the problems of protecting users and financial infrastructure brought by the increasing popularity of mobile payment platforms. This project explores various security threats faced by existing mobile payment methods and plans to combine zero-trust architecture security solutions in financial infrastructure. Sub-project 6: We will develop a policy-checking mechanism in the fully homomorphic encryption evaluation function and construct a machine unlearning technique without revoking user data. Sub-project 7: Design a privacy-preserving data search mechanism suitable for fintech with quantum-resistant security and a flexible fine-grained access control mechanism.

It is expected that through this project, a university-level information security center will be established at NCCU to enhance the country’s information security capabilities, expand international cooperation, and cultivate future national information security talents.