IMDEA Networks
Tianyue CHU
PhD Student
- Affiliation: IMDEA Networks Institute, Universidad Carlos III de Madrid (U3CM)
- BSc: Double Bachelor's Degree. Mathematics and Applied Mathematics & Finance - Shenzhen University. Shenzhen, China
- MSc: Statistics - Shenzhen University. Shenzhen, China
- Former position: Research Assistant. Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences. Shenzhen. China
- Joining date: October 2020
- Email: tianyue.chu@imdea.org
- Phone: +34 91 481 6978
Publications
Information-Theoretical Bounds on Privacy Leakage in Pruned Federated Learning
Tianyue Chu, Mengwei Yang, Nikolaos Laoutaris, Athina Markopoulou
ISIT 2024 Workshop on Information-Theoretic Methods for Trustworthy Machine Learning. Athens, Greece. July 2024
FedQV: Leveraging Quadratic Voting in Federated Learning
Tianyue Chu, Nikolaos Laoutaris
ACM SIGMETRICS. Venice, Italy. June 2024
Strengthening Privacy in Robust Federated Learning through Secure Aggregation
Tianyue Chu, Devriş İşler, Nikolaos Laoutaris
Workshop on Artificial Intelligence System with Confidential Computing (AISCC 2024), co-located with NDSS Symposium 2024. San Diego, CA, USA. February 2024
Securing Federated Sensitive Topic Classification against Poisoning Attacks
Tianyue Chu, Álvaro García-Recuero, Costas Iordanou, Georgios Smaragdakis, Nikolaos Laoutaris
Usenix Network and Distributed System Security Symposium. San Diego, California. February 2023
-
Pre-impact alarm system for fall detection using MEMS sensors and HMM-based SVM classifiercall_made
.
Accidental fall can cause physical injury, fracture and other health complications, especially for elderly people living alone. Aimed to provide timely assistance after the occurrence of falling down, a pre-fall alarm system was proposed. In order to test the reliability of the pre-fall alarm system, eighteen subjects who wore this device on the waist were required to participate in a series of experiments. The acceleration and angular velocity time series extracted from human motion processes were used to describe human motion features. HMM-based SVM classifier was used to determine the maximum separation boundary between falls and Activities of Daily Living (ADLs). The fall detection results showed 94.91% accuracy, 97.22% Sensitivity and 93.75% Specificity. The proposed device can accurately recognize fall events, achieve additional functions, and have the advantages of small size and low power consumption. Based on the findings, this pre-impact fall alarm system with a detection algorithm could potentially be useful for monitoring the state of physical function in elderly population.