About Me

I’m Xianqi Deng, a second-year PhD student at the University at Albany (SUNY). My research explores how Large Language Models can accelerate scientific discovery. Lately, I’ve focused on masked language modeling for molecular representation learning—teaching models to capture structure–function relationships that improve prediction and design tasks. Broadly, I’m interested in self-supervised learning, generative modeling, and practical benchmarks that connect model capability to real scientific impact.

Research

1. Domain Knowledge Informed Parameter-Efficient Fine-Tuning
2. Learnable Masking for Masked Language Modeling in Scientific Area
3. Generative Models for Nanomaterial Design and discovery
4. Molecular Structures Derivation from IR Spectra with Seq2Seq Model

Publications

# First or co-first author.

* Corresponding author.

2026

1. Siqi Chen#, Zhiqiang Wang#, Yili Shen, Xianqi Deng, Xi Cheng, Cheng-Wei Ju, Jun Yi, Guo Ling, Dieaa Alhmoud, Hui Guan*, Zhou Lin*
   Transferable FB-GNN-MBE Framework for Potential Energy Surfaces: Data-Adaptive Transfer Learning in Deep Learned Many-Body Expansion Theory.
   Under review at The Journal of Chemical Physics. arXiv: 2604.09320
2. Elham Sadeghi#, I-Hsin Lin, Xianqi Deng, Stacy M Copp*, Petko Bogdanov*
   Geometry-Preserving Supervised Biological Sequence Design.
   KDD 2026 AI4Sciences Track Poster.

2025

1. Ethan French#, Xianqi Deng, Siqi Chen, Cheng-Wei Ju, Xi Cheng, Lijun Zhang, Xiao Liu, Hui Guan*, Zhou Lin*
   Revolutionizing Spectroscopic Analysis Using Sequence-to-Sequence Models I: From Infrared Spectra to Molecular Structures.
   Under revision at Journal of the American Chemical Society. ChemRxiv: 2025-n4q84
2. Elham Sadeghi#, Xianqi Deng, I-Hsin Lin, Stacy M Copp*, Petko Bogdanov*
   Property-Isometric Variational Autoencoders for Sequence Modeling and Design.
   arXiv: 2509.14287

2024

1. Siqi Chen#, Zhiqiang Wang#, Xianqi Deng#, Yili Shen#, Cheng-Wei Ju, Jun Yi, Lin Xiong, Guo Ling, Dieaa Alhmoud, Hui Guan*, Zhou Lin*
   Integrating Graph Neural Networks and Many-Body Expansion Theory for Potential Energy Surfaces.
   NeurIPS 2024 AI4Mat Workshop Spotlight. arXiv: 2411.01578

Service

• The 14th IEEE International Conference on Healthcare Informatics (IEEEICHI2026), Reviewer.