Outstanding Computer Science Undergraduates Shine at ASMPT Technology Award 2025

Two outstanding Computer Science undergraduates, Kaitai Zhang and Fengfei Yu, achieved a great success at the ASMPT Technology Award 2025, held on 27 June. Kaitai Zhang secured the Silver Award, while Fengfei Yu received the Outstanding Award for their respective final year projects.

Kaitai Zhang’s award-winning project, titled “Dominating MOBA Games with VLL: An Efficient Gaming AI System,” introduces Visual-LLM-Logic (VLL), a hybrid AI framework that comprised three modular layers. The perception layer employs YOLO for real-time object detection, the strategic reasoning layer leverages a large language model for high-level planning, and the logic-based decision layer executes tactical actions. This asynchronous structure facilitates effective collaboration among components, enabling rapid adaptation and transparent decision-making. By integrating visual perception with language-driven reasoning, VLL overcomes traditional limitations of game AI, resulting in a more adaptive and context-aware agent.

Fengfei Yu’s Outstanding Award project, titled “Defending Against Model Inversion via Auxiliary-Supervised OOD Regularisation,” analyses the leakage of training data through model inversion attacks. It highlights a critical flaw in existing defences, which often assign excessive confidence to out-of-distribution (OOD) inputs, allowing inversion attacks that reconstruct training data. To address this, the Fengfei proposes Dual Confidence Calibration (DCC), a novel defence mechanism that enhances detection of both in-distribution and OOD samples. Empirical evaluations demonstrate that DCC significantly reduces attack success rates, confirming its effectiveness as a defensive solution.

Upon receiving the award, Kaitai Zhang expressed his gratitude to his project supervisor, Professor Renjie Wan, for his invaluable guidance, stating, “This journey has deepened my understanding of advanced technologies and strengthened my problem-solving skills.” Fengfei Yu also shared his appreciation, “I am deeply thankful to my supervisor, Professor Bo Han, for his unwavering support. This award has been a catalyst for my professional growth, challenging me to bridge technical research with tangible social impact and enhancing my understanding of how innovation should ultimately serve society.”

Since its inception in 2015, the ASMPT Technology Award, organised by ASMPT Limited, has served as a prestigious platform for engineering students to showcase their research achievements. This year, 12 outstanding projects were nominated to compete for awards, evaluated by a panel of judges comprising ASMPT professionals, university scholars, and industry experts.