Publications
My research spans physics-informed machine learning and human-computer interaction. Below are my published and in-progress works.
Published
Physics-Informed Neural Networks for Scientific Computing
IEEE Xplore · CSCI 2024 · Published
This paper presents a physics-informed neural network framework that integrates governing partial differential equations directly into the neural network training process. We propose enhanced architectures using Fourier feature embeddings to address spectral bias and adaptive loss weighting to balance competing PDE constraints. Validated on benchmark equations including Burgers’ equation and heat diffusion, the approach demonstrates improved convergence and accuracy over standard PINN implementations.
My Contribution: Designed and implemented the PINN architectures, conducted all experiments, and led the manuscript writing.
Keywords: Physics-Informed Neural Networks, PDE, Fourier Features, Scientific Computing, Deep Learning
Under Review
Affective Modeling in AI Narrative Systems
ACM CHI 2026 · Under Review
This work introduces an emotion-aware AI narrative system that dynamically adapts storytelling based on real-time user affect. We benchmark six deep learning architectures (LSTM, CNN, Transformer, ResNet, InceptionTime, LSTM-FCN) for physiological time series classification and apply Grad-CAM for temporal interpretability. The system demonstrates how affective computing can create more responsive and empathetic interactive narrative experiences, with implications for therapeutic and educational applications.
My Contribution: Contributing author — implemented deep learning benchmarks, developed Grad-CAM temporal visualizations, and contributed to experimental design and statistical analysis.
Keywords: Affective Computing, AI Narratives, Time Series Classification, Grad-CAM, Human-Computer Interaction
Xuyang Chen 