Mingjie Zhu

Mingjie Zhu, Yejian Lyu, Ziming Yu et al.

Terahertz (THz) wireless communication has emerged as a promising solution for future data center interconnects; however, accurate channel characterization and system-level performance evaluation in complex indoor environments remain challenging. In this work, a measurement-calibrated AI-assisted digital twin (DT) framework is developed for THz wireless data centers by tightly integrating channel measurements, ray-tracing (RT), and implicit neural field (INF) modeling. Specifically, channel measurements are first conducted using a vector network analyzer at 300 GHz under both line-of-sight (LoS) and non-line-of-sight (NLoS) scenarios. RT simulations performed on the Sionna platform capture the dominant multipath structures and show good consistency with measured results. Building upon measurement and RT data, an RT-conditioned INF is developed to construct a continuous radio-frequency (RF) field representation, enabling accurate prediction in RT-missing NLoS regions. The comprehensive RF map generated by DT can provide system-level analysis and decisions for wireless data centers.

Yuxing Tian, Mingjie Zhu, Jiachi Luo et al.

This study focuses on long-term forecasting (LTF) on continuous-time dynamic graph networks (CTDGNs), which is important for real-world modeling. Existing CTDGNs are effective for modeling temporal graph data due to their ability to capture complex temporal dependencies but perform poorly on LTF due to the substantial requirement for historical data, which is not practical in most cases. To relieve this problem, a most intuitive way is data augmentation. In this study, we propose \textbf{\underline{U}ncertainty \underline{M}asked \underline{M}ix\underline{U}p (UmmU)}: a plug-and-play module that conducts uncertainty estimation to introduce uncertainty into the embedding of intermediate layer of CTDGNs, and perform masked mixup to further enhance the uncertainty of the embedding to make it generalize to more situations. UmmU can be easily inserted into arbitrary CTDGNs without increasing the number of parameters. We conduct comprehensive experiments on three real-world dynamic graph datasets, the results demonstrate that UmmU can effectively improve the long-term forecasting performance for CTDGNs.

Fei Zheng, Chaochao Chen, Xiaolin Zheng et al.

With the increasing demands for privacy protection, privacy-preserving machine learning has been drawing much attention in both academia and industry. However, most existing methods have their limitations in practical applications. On the one hand, although most cryptographic methods are provable secure, they bring heavy computation and communication. On the other hand, the security of many relatively efficient private methods (e.g., federated learning and split learning) is being questioned, since they are non-provable secure. Inspired by previous work on privacy-preserving machine learning, we build a privacy-preserving machine learning framework by combining random permutation and arithmetic secret sharing via our compute-after-permutation technique. Since our method reduces the cost for element-wise function computation, it is more efficient than existing cryptographic methods. Moreover, by adopting distance correlation as a metric for privacy leakage, we demonstrate that our method is more secure than previous non-provable secure methods. Overall, our proposal achieves a good balance between security and efficiency. Experimental results show that our method not only is up to 6x faster and reduces up to 85% network traffic compared with state-of-the-art cryptographic methods, but also leaks less privacy during the training process compared with non-provable secure methods.