I am a Master’s student in Cyberspace Security at Xidian University, conducting research at the Faculty of Cybersecurity & Cryptology under the supervision of Prof. Hao Wang. Previously, I earned a Bachelor’s degree in Computer Science and Technology from Tianjin University of Technology.

My research interests primarily lie in diffusion model acceleration, multimodal cardiac diagnosis, and neuro-symbolic reasoning. I am particularly interested in leveraging advanced architectures like Mamba (State-Space Models) for processing long-sequence physiological signals, and exploring statistically optimal acceleration techniques for large generative models. Currently, I am leading the Cardio-Verse (心视界) project, an AI-driven multimodal cardiac diagnostic platform.

I have published and presented my research at top international academic venues including ICML, ICASSP, ICME, and BIBM.

🔥 News

• 2026.05:  🎉🎉 Our two papers were accepted at ICML 2026.
• 2026.05:  🎉🎉 I will be traveling to Barcelona, Spain to present my first-author papers at ICASSP 2026.
• 2026.03:  🎉🎉 Two papers on Mamba Diffusion were accepted at ICME 2026.
• 2026.01:  🚀🚀 Officially launched the Cardio-Verse (心视界) multimodal cardiac diagnostic platform project.
• 2026.01:  🎉🎉 Three first-author papers were accepted at ICASSP 2026.
• 2025.12:  🎉🎉 Our paper on Hierarchical Latent Diffusion Models (H-LDM) for PCG synthesis was accepted at BIBM 2025.

📝 Publications

Accepted

Cardiac murmur grading and risk analysis of cardiac diseases based on adaptable heterogeneous-modality multi-task learning

Chenyang Xu, Xin Li, Xinyue Zhang, et al.

Health Information Science and Systems, 2024

PDF | Code

Accepted

Enhancing Respiratory Sound Detection Through Integrated Learning and Dynamic Convolutional Recurrent Neural Networks

Chenyang Xu, Yuanbing Ouyang, Ziye Chen, Hao Wang.

BioCAS 2024

PDF | Code

Accepted

H-LDM: Hierarchical Latent Diffusion Models for Controllable and Interpretable PCG Synthesis from Clinical Metadata

Chenyang Xu, Siming Li, Hao Wang.

BIBM 2025 (Oral Presentation)

PDF | Code

Accepted

Disentangling Physiology from Fidelity: Latent-Guided Diffusion Models for Cross-Modal Cardiac Synthesis

Chenyang Xu, Siming Li, Wensai Xuan, Hao Wang.

ICASSP 2026

PDF | Code

Accepted

AdaGrad-Fusion: Adaptive Gradient Fusion for Memory-Efficient ECG Foundation Model Fine-Tuning

Chenyang Xu, Yi Zhao, Zezhou Zhang, Hao Wang.

ICASSP 2026

PDF | Code

Accepted

CardioBridge-DM: Bridging Cross-Cohort Heart Sound Synthesis via Rhythm-Aware Semi-Supervised Diffusion

Chenyang Xu, Siming Li, Hao Wang.

ICASSP 2026

PDF | Code

Accepted

PhysioMamba: Physics-Informed ST-Mamba Diffusion for Controllable Echocardiogram Synthesis

Chenyang Xu, Dezhen Wang, Lin Chen, and Hao Wang.

Xidian University, Qingdao University of Technology, Beijing Technology and Business University

ICME 2026

PDF | Code

• Proposed a novel physics-informed spatio-temporal Mamba diffusion model designed to synthesize high-fidelity, controllable echocardiogram videos, accelerating the generation process while maintaining physiological accuracy.

Accepted

CardioManifold: Direct Pathological PCG Synthesis via Cyclic-Mamba and Phase-Preserving Manifold Diffusion

Chenyang Xu, Dezhen Wang, and Hao Wang.

Xidian University, Qingdao University of Technology

ICME 2026

PDF | Code

Accepted

EigenCache: Rethinking Diffusion Acceleration as Covariance-Optimal Forecasting and Submodular Information Allocation

Chenyang Xu, Dezhen Wang, Lin Chen, et al.

ICML 2026

PDF | Code

• Proposed EigenCache, a theoretically grounded framework that re-frames diffusion acceleration as a problem of covariance learning and optimal experimental design. It replaces rigid polynomial bases with a covariance-optimal forecaster (Kriging) and selects anchor timesteps via a provably near-optimal submodular information schedule, establishing a new Pareto frontier for efficient inference.

Accepted

Unlocking Cross-Modal Biosignal Synthesis: A Temporally-Aware VAE-Diffusion Model

Chenyang Xu, Dezhen Wang, Hao Wang.

ICML 2026

PDF | Code

🎖 Honors and Awards

• 2024 Second Place, IEEE BioCAS Grand Challenge
• 2025 IEEE BIBM Student Travel Grant
• 202x Gold Award, “Internet+” Innovation and Entrepreneurship Competition (Tianjin)
• 202x Grand Prize, Group Programming Ladder Tournament (GPLT) (Tianjin)
• 202x First Prize, Lanqiao Cup National Software and Information Technology Professionals Competition (Tianjin)
• 202x Bronze Award, “Challenge Cup” National College Student Extracurricular Academic Science and Technology Works Competition (Tianjin)
• 202x Third Prize, 4th China Postgraduate FinTech Innovation Competition
• 2024 Outstanding Graduate Student, Xidian University
• 2024 First-class Graduate Academic Scholarship, Xidian University
• 2021-2023 First-class People’s Scholarship (Three Consecutive Years), Tianjin University of Technology

📖 Education

• 2024.09 - Present, M.Eng. in Cyberspace Security, Xidian University (Faculty of Cybersecurity & Cryptology).
• 2020.09 - 2024.06, B.Eng. in Computer Science and Technology, Tianjin University of Technology.

💻 Projects & Experience

• 2024.12 - Present, Project Lead, Large Model Inference Optimization in Resource-Constrained Environments (Ant Group & School of Cyber Engineering Innovation Project)
• 2024.09 - Present, Project Lead, Cardio-Verse (心视界)
  • Leading the research and development of a multimodal cardiac diagnostic platform. Responsible for algorithm design and multimodal data integration.
• 2023.02 - 2024.09, Research Assistant, National Institute of Health Data Science at Peking University