Xiang Shaowen

Xiang Shaowen

Ph.D. Student @ UMich ECE
DeepThink Lab & PixAIL

Posts by Collection

portfolio

publications

DuetUI: A Bidirectional Context Loop for Human-Agent Co-Generation of Task-Oriented Interfaces

Published in ACM Conference on Human Factors in Computing Systems (CHI), 2025

A bidirectional context loop enabling human–agent co-generation of task-oriented interfaces. Accepted at ACM CHI.

Recommended citation: Y. Xu, S. Xiang, Y. Song, R. Sun, and X. Tong. (2025). "DuetUI: A Bidirectional Context Loop for Human-Agent Co-Generation of Task-Oriented Interfaces." ACM CHI. (accepted). arXiv:2509.13444.
Download Paper

OptiMA: Collaborative Multi-Agent Framework for Modelling and Controlling Raman Amplifier in Intelligent Optical Networks

Published in European Conference on Optical Communication (ECOC), Copenhagen, Denmark, 2025

A multi-agent LLM framework for intelligent optical networks, integrating real-time control, digital-twin modeling, and pump optimization for Raman amplifiers.

Recommended citation: S. Xiang, S. Wu, X. Liu, Q. Qiu, Y. Zhang, Y. Chen, and Q. Zhuge. (2025). "OptiMA: Collaborative Multi-Agent Framework for Modelling and Controlling Raman Amplifier in Intelligent Optical Networks." ECOC 2025, Copenhagen, Denmark. IEEE Xplore.
Download Paper

Field-trial Investigations into the Impact of Digital Twin Accuracy on Optical Power Optimization

Published in Asia Communications and Photonics Conference (ACP) 2025, 2025

Field-trial study on how digital twin accuracy affects optical power optimization in deployed optical networks. Accepted at ACP 2025.

Recommended citation: X. Liu, Y. Cheng, S. Xiang, M. Tornatore, Q. Qiu, Y. Zhang, W. Hu, and Q. Zhuge. (2025). "Field-trial Investigations into the Impact of Digital Twin Accuracy on Optical Power Optimization." ACP 2025. IEEE Xplore.
Download Paper

AI-generated data contamination erodes pathological variability and diagnostic reliability

Published in arXiv preprint, 2026

Large-scale study of how AI-generated synthetic data contamination degrades pathological variability and diagnostic reliability in medical AI. arXiv:2601.12946.

Recommended citation: H. He, S. Xiang, Y. Zhang, Y. Zhu, J. Zhang, H. Deng, E. Alsentzer, Q. Chen, K.-H. Yu, A. Marmenshall, T. Chen, S. Anumasa, D. Ebner, D. Ho, K. Y. Ngiam, C.-Y. Cheng, and D. Liu. (2026). "AI-generated data contamination erodes pathological variability and diagnostic reliability." arXiv:2601.12946.
Download Paper

Reliable QoT Estimation for Imbalanced Data from Deployed Optical Networks Using MDN-ResKAN

Published in Asia Communications and Photonics Conference (ACP) 2026 — under review, 2026

MDN-ResKAN models Q-factor distributions for reliable QoT estimation under imbalanced deployed-network data. Under review at ACP 2026.

Recommended citation: S. Xiang, S. Wu, Q. Qiu, R. Liu, Y. Zhang, W. Hu, and Q. Zhuge. (2026). "Reliable QoT Estimation for Imbalanced Data from Deployed Optical Networks Using MDN-ResKAN." Submitted to ACP 2026. (under review).
Download Paper

A Physics-Guided Learning Framework for Accurate and Generalizable EDFA Gain Spectrum Modeling with FourierKAN

Published in Journal of Lightwave Technology (JLT) — under review [Invited], 2026

Physics-guided FourierKAN for EDFA gain spectrum modeling with strong generalization under unseen configurations. Under review at Journal of Lightwave Technology (invited).

Recommended citation: Q. Qiu, Y. Zhu, S. Xiang, Y. Zhang, J. Wen, Z. Zhang, W. Hu, and Q. Zhuge. (2026). "A Physics-Guided Learning Framework for Accurate and Generalizable EDFA Gain Spectrum Modeling with FourierKAN." Submitted to Journal of Lightwave Technology. (under review, invited).
Download Paper

talks

teaching