Publications

My reasearch is at the intersection of multimodal learning, large language models, and biomedical AI.
I build interpretable multimodal mixture-of-experts (MoE) systems, evaluate large language models, and apply machine learning to proteins, mass spectrometry, and whole-slide images.

Quick Navigation

• Multimodal Representation & MoE
• Large Language Models Evals & Reasoning
• Biomedical AI & Computational Biology

Multimodal Representation & MoE

* denotes equal contribution

Thumbnail	Details
	I2MoE: Interpretable Multimodal Interaction-Aware Mixture-of-Experts Xin, J., Yun, S., Peng, J., Choi, I., Ballard, J. L., Chen, T., Long, Q. ICML 2025 We introduce a drop-in MoE framework that disentangles redundancy, synergy, and uniqueness interactions between different modalities, achieving superior multimodal fusion performance while exposing interaction weights for user interpretation. Paper
	Generate, Then Retrieve: Addressing Missing Modalities in Multimodal Learning via Generative AI and MoE Yun, S.*, Xin, J.*, Choi, I., Peng, J., Ding, Y., Long, Q., Chen, T. AAAI GenAI4Health Workshop 2025 Best Paper We impute missing modalities with learnable embeddings and route them through a sparse MoE, outperforming strong baselines on real-world multimodal datasets. Paper
	Modalities Contribute Unequally: Enhancing Medical Multi-modal Learning through Adaptive Modality Token Re-balancing Peng, J.*, Ballard, J. L*,., Zhang, M., Yun, S., Xin, J., Long, Q., Zhang, Y., Chen, T. ICML 2025 We propose Adaptive Modality Token Re-BalanCing (AMC), a novel top-down dynamic multi-modal fusion approach. Paper
	Flex-MoE: Modeling Arbitrary Modality Combinations via the Flexible Mixture-of-Experts Yun, S., Choi, I., Peng, J., Wu, Y., Bao, J., Zhang, Q., Xin, J., Long, Q., Chen, T. NeurIPS 2024 Spotlight We propose Flex-MoE, new MoE framework designed to flexibly incorporate arbitrary modality combinations while maintaining robustness to missing data. Paper

Large Language Models Evals & Reasoning

Thumbnail	Details
	Evaluating the Unseen Capabilities: How Many Theorems Do LLMs Know? Li, X., Xin, J., Long, Q., Su, W. Under review 2025 We introduce KnowSum, a statistical framework designed to provide a more comprehensive assessment by quantifying the unseen knowledge for a class of evaluation tasks. Paper
	Selective Annotation Makes Language Models Better Few-Shot Learners Su, H., Kasai, J., Wu, C.-H., Shi, W., Wang, T., Xin, J., Zhang, R., Ostendorf, M., Zettlemoyer, L., Smith, N. A., Yu, T. ICLR 2023 We formulate an annotation-efficient, two-step framework: selective annotation that chooses a pool of examples to annotate fromunlabeled data in advance, followed by prompt retrieval that retrieves task examples from the annotated pool at test time. Paper
Upcoming in August	One Paper about LLM Reasoning

Biomedical AI & Computational Biology

Thumbnail	Details
	TAD-Graph: Enhancing Whole-Slide Image Analysis via Task-Aware Subgraph Disentanglement Wang, F., Xin, J., Zhao, W., Jiang, Y., Yeung, M., Wang, L., Yu, L. IEEE TMI 2025 We propose a novel Task-Aware Disentanglement Graph approach that operates on WSI graph representations, effectively identifying and disentangling informative subgraphs to enhance contextual feature extraction. Paper
	Retrieved Sequence Augmentation for Protein Representation Learning Ma, C., Zhao, H., Zheng, L., Xin, J., Li, Q., Wu, L., Deng, Z., Lu, Y. Y., Liu, Q., Wang, S., Kong, L. EMNLP 2024 We show that a simple alternative, Retrieved Sequence Augmentation (RSA), can enhance protein representation learning without the need for alignment and cumbersome preprocessing. Paper
	MIST-CF: Chemical Formula Inference from Tandem Mass Spectra Goldman, S.*, Xin, J.*, Provenzano, J., Coley, C. W. Journal of Chemical Information and Modeling 2023 We extend previous spectrum Transformer methodology for learning to rank chemical formula and adduct assignments given an unannotated tandem MS spectrum. Paper
	Prefix-Tree Decoding for Predicting Mass Spectra from Molecules Goldman, S., Bradshaw, J., Xin, J., Coley, C. W. NeurIPS 2023 Spotlight We use a new intermediate strategy for predicting mass spectra from molecules by treating mass spectra as sets of molecular formulae, which are themselves multisets of atoms. Paper
	Artificial intelligence clinical applications of wearable technologies Ma, S., Yee, C. Y., Xin, J., Ho, J. W. K. Book chapter in Machine Learning, Medical AI and Robotics (IOP) 2023 We survey deep learning-based AI applications of wearable devices. Book Chapter