Publications

Selected work. * represents equal contribution.

Latest work

Preprints

1 paper
Abstract visual for CAPT: A Multi-task Continuous Autoregressive Transformer Enabling Cross-dataset and Cross-species Transfer for Calcium Population Dynamics

CAPT: A Multi-task Continuous Autoregressive Transformer Enabling Cross-dataset and Cross-species Transfer for Calcium Population Dynamics

Xinhong Xu*, Yimeng Zhang*, Yuanlong Zhang

arXiv preprint, 2026

CAPT directly models continuous calcium traces and evaluates frozen-backbone transfer across independent mouse, zebrafish, and C. elegans datasets.

PaperCode
Peer reviewed

Conference Papers

2 papers
Abstract visual for CalM: A Self-Supervised Foundation Model for Calcium-imaging Population Dynamics

CalM: A Self-Supervised Foundation Model for Calcium-imaging Population Dynamics

Xinhong Xu*, Yimeng Zhang*, Qichen Qian, Yuanlong Zhang

ICML 2026

CalM learns reusable representations from calcium traces for neural forecasting, behavior decoding, and functional analysis.

PaperCode
Abstract visual for AdaFocusV3: On Unified Spatial-temporal Dynamic Video Recognition

AdaFocusV3: On Unified Spatial-temporal Dynamic Video Recognition

Yulin Wang*, Yang Yue*, Xinhong Xu, Ali Hassani, Victor Kulikov, Nikita Orlov, Shiji Song, Humphrey Shi, Gao Huang

ECCV 2022

AdaFocusV3 reduces spatial and temporal redundancy by allocating computation to informative video cubes.

PaperCode
Peer reviewed

Journal Articles

2 papers
Abstract visual for Constructing Mesoscale Functionomics by Neural Dynamics Subspace Clustering

Constructing Mesoscale Functionomics by Neural Dynamics Subspace Clustering

Yeyi Cai*, Xinhong Xu*, Guihua Xiao, Huaming Ling, Zuoqiang Shi, Minghuan Wang, Mingrui Wang, Fenghua Chen, Jiamin Wu, Qionghai Dai

National Science Review, 2026

FRID identifies functional neural populations through intrinsic dynamics rather than correlated firing alone.

Paper
Abstract visual for Unsupervised Transfer Learning Enables Multi-animal Tracking without Training Annotation

Unsupervised Transfer Learning Enables Multi-animal Tracking without Training Annotation

Yixin Li, Qi Zhang, Yuanlong Zhang, Jiaqi Fan, Zhi Lu, Xinhong Xu, Xinyang Li, Ziwei Li, Jiamin Wu, Qionghai Dai

Nature Methods, 2026

UDMT performs annotation-free multi-animal tracking under crowding, occlusion, rapid motion, low contrast, and cross-species settings.

PaperCode