Shaobo Han

Senior Researcher

NEC Labs America

Biography

I am a Senior Researcher at NEC Laboratories America, where I work on the design and development of machine learning and signal processing algorithmic solutions for real-world sensing applications. I’m interested in generalizable AI methods that can address limitations in training data acquisition and domain shifts in deployment environments. I received my Ph.D. in Electrical and Computer Engineering and M.S. in Statistical Science from Duke University. My Ph.D. advisor was Lawrence Carin. I am a Senior Member of IEEE.

His research interest includes:

Probabilistic/deep generative models and structured variational inference,
Multitask/transfer learning, cross-domain model adaptation,
Bayesian statistics, sparsity, and latent variable models,
Signal processing and sensing applications.

News

[2024/12] Two papers were accepted to ICASSP 2025, and four papers were accepted to OFC 2025. The topics include (i) disaggregated inference and fine-tuning over optical networks, and (ii) the adaptation of large-scale pre-trained audio language models to downstream tasks.
[2024/09] Our paper, “VB-LoRA: Extreme Parameter Efficient Fine-Tuning with Vector Banks”, has been accepted at NeurIPS 2024. The code has been merged into the Hugging Face PEFT library.
[2024/05] Our paper “Deep learning-based intrusion detection and impulsive event classification for distributed acoustic sensing across telecom networks” has been accepted by Journal of Lightwave Technology.
[2023/08] UAI 2023 Top Reviewer.
[2023/05] Our work on fiber sensing AI was featured by Laser Focus World: Top Stories and Photonics Hot List.
[2023/05] Guest Lecture: Automatic Differentiation and Differentiable Programming, In CS 610: Data Structure and Algorithms, NJIT.
[2022/09] Our paper Using Global Existing Fiber Networks for Environmental Sensing" has been accepted by Proceedings of the IEEE.
[2022/07] Received Outstanding Performance Award 2022, Global Innovation Unit, NEC Corporation.
[2022/01] Two papers accepted to ICLR 2022. Topics include (i) test-time policy adaptation under ordinal reward, and (ii) zero-shot domain adaptation with multiway categorical domain descriptor.
[2021/06] Two AI-based fiber sensing solutions have been commercialized: NEC’s press release.

Selected Publications

CLAP-S: Support Set-Based Adaption for Downstream Fiber-Optic Acoustic Recognition,
Jingchen Sun, Shaobo Han, Wataru Kohno, and Changyou Chen.
IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2025), Lecture,
[Link] [PDF] [GitHub] [Gunshot-Firework Dataset]
Adaptation of multimodal foundation models to real-world downstream tasks, under severe domain shifts
VB-LoRA: Extreme Parameter Efficient Fine-Tuning with Vector Banks,
Yang Li, Shaobo Han, and Shihao Ji.
Neural Information Processing Systems (NeurIPS 2024),
[PDF] [GitHub] [Hugging Face PEFT]
Learned global parameter sharing paradigm, differentiable Top-K
Deep Learning-Based Intrusion Detection and Impulsive Event Classification for Distributed Acoustic Sensing Across Telecom Networks,
Shaobo Han, Ming-Fang Huang, Tingfeng Li, Jian Fang, Zhuocheng Jiang, and Ting Wang.
Journal of Lightwave Technology (JLT) , Vol. 42, No.12 (2024) 4167-4176,
[Link] [PDF] [Gunshot-Firework Dataset]
Learning Transferable Reward for Query Object Localization with Policy Adaptation,
Tingfeng Li, Shaobo Han, Martin Renqiang Min and Dimitris N. Metaxas.
International Conference on Learning Representations (ICLR 2022),
[PDF] [OpenReview] [GitHub]
Test-time adaptation, extrapolation from one-class
Provable Adaptation across Multiway Domains via Representation Learning,
Zhili Feng, Shaobo Han and Simon S. Du.
International Conference on Learning Representations (ICLR 2022),
[PDF] [OpenReview]
Zero-shot domain adaptation, compositional generalization
Automatic Fine-grained Localization of Utility Pole Landmarks on Distributed Acoustic Sensing Traces based on Bilinear ResNets,
You Lu, Yue Tian, Shaobo Han, Eric Cosatto, Sarper Ozharar, and Yangmin Ding.
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2021),
[Link]
VAE Learning via Stein Variational Gradient Descent,
Yunchen Pu, Zhe Gan, Ricardo Henao, Chunyuan Li, Shaobo Han and Lawrence Carin.
Neural Information Processing Systems (NIPS 2017),
[PDF] [Link]
Variational Gaussian Copula Inference,
Shaobo Han, Xuejun Liao, David B. Dunson, and Lawrence Carin.
International Conference on Artificial Intelligence and Statistics (AISTATS 2016),
[PDF] [Supplementary] [GitHub]
Alternating Minimization Algorithm with Automatic Relevance Determination for Transmission Tomography under Poisson Noise,
Yan Kaganovsky, Shaobo Han, Soysal Degirmenci, David G. Politte, David J. Brady, Joseph A. O'Sullivan, and Lawrence Carin.
SIAM Journal on Imaging Sciences (SIIMS), Vol. 8, No.3 (2015) 2087-2132
[PDF] [GitHub]
Dynamic Rank Factor Model for Text Streams,
Shaobo Han, Lin Du, Esther Salazar, and Lawrence Carin.
Neural Information Processing Systems (NIPS 2014),
[PDF] [Supplementary] [Link]
Hierarchical Infinite Divisibility for Multiscale Shrinkage,
Xin Yuan, Vinayak Rao, Shaobo Han, and Lawrence Carin.
IEEE Transactions on Signal Processing (IEEE-TSP), Vol. 62, No.17 (2014) 4363-4374
[PDF] [Supplementary] [Code]
Integrated Non-Factorized Variational Inference,
Shaobo Han, Xuejun Liao, and Lawrence Carin.
Neural Information Processing Systems (NIPS 2013),
[PDF] [Supplementary] [Link]
Cross-Domain Multitask Learning with Latent Probit Models,
Shaobo Han, Xuejun Liao, and Lawrence Carin.
International Conference on Machine Learning (ICML 2012),
[PDF]