|
I am currently a Shanahan Foundation Fellow at the Allen Institute, and postdoctoral researcher at the University of Washington, Seattle, where I closely work with Prof. Matthew Golub, Dr. Uygar Sümbül, Prof. Edgar Walker, and Prof. Eric Shea-Brown. Prior to that, I received my Ph.D. at MIT CSAIL in 2022, where I studied in Computational Connectomics Group. I was fortunate to be advised by Prof. Nir Shavit at MIT, and co-advised by Prof. Aravinthan D.T. Samuel and Prof. Jeff W. Lichtman at Harvard Center for Brain Science. Meanwhile, I was a visiting researcher at HHMI Janelia Research Campus working with Dr. Srinivas C. Turaga. I also held research internship at Google, Waymo. I received my M.S. degree at MIT EECS in 2019, and B.S. degree at Tsinghua in 2017. Email / Google Scholar / Twitter / CV |
|
|
My research interest is NeuroAI, which studies the intersection of natural intelligence and artificial intelligence. During my PhD, I developed advanced deep learning tools for fast and scalable automatic connectomics pipelines to discover the brain, and linking the anatomical structure and neural activity with whole-brain modeling.
In the current collaboration with investigators at the Allen Institute and the University of Washington, I am developing interpretable deep learning approaches including probabilistic modeling and representation learning for biological and artificial systems. My research topics broadly include
|
|
|
|
| * indicates equal contribution | |
|
Yicong Li, Yaron Meirovitch, Aaron T. Kuan, Jasper S. Phelps, Alexandra Pacureanu, Wei-Chung Allen Lee, Nir Shavit, Lu Mi IEEE - ISBI 2023 We propose an uncertainty-aware 3D reconstruction model that translates X-Ray images to EM-like images with enhanced membrane segmentation quality. |
|
Tri Nguyen, Mukul Narwani, Mark Larson, Yicong Li, Shuhan Xie, Hanspeter Pfister, Donglai Wei, Nir Shavit, Lu Mi, Alexandra Pacureanu, Wei-Chung Lee, Aaron T. Kuan IEEE - ISBI Challenge 2023 We organize an instance segmentation challenge with volumetric XNH images of cortical white matter axons from the mouse brain at 100 nm per voxel isotropic resolution. Both skeleton and voxel ground truth annotations are provided for axon trajectories. |
|
Lu Mi, Abhijit Kundu, David Ross, Frank Dellaert, Noah Snavely, Alireza Fathi We propose im2nerf, a learning framework with deep generative model that predicts a continuous neural object representation given a single input image in the wild, supervised by only segmentation output from off-the-shelf recognition methods. |
|
Lu Mi, Richard Xu, Sridhama Prakhya, Albert Lin, Nir Shavit, Aravinthan D.T. Samuel, Srinivas C. Turaga ICLR 2022 We use stochastic threshold linear dynamics to model the whole-brain neural network, using connectome constraint on the neural activity. |
|
Lu Mi, Hao Wang, Yonglong Tian, Hao He, Nir Shavit AAAI 2022 ICML 2021 Workshop on Uncertainty & Robustness in Deep Learning We perform a systematic exploration into training-free uncertainty estimation for dense regression, an unrecognized yet important problem, and provide a theoretical construction justifying such estimations. |
|
Lu Mi, Hang Zhao, Charlie Nash, Xiaohan Jin, Jiyang Gao, Chen Sun, Cordelia Schmid, Nir Shavit, Yuning Chai, Dragomir Anguelov CVPR 2021 We propose HDMapGen, a hierarchical graph generative model capable of producing high-quality and diverse HD maps. |
|
Lu Mi, Richard Xu, Sridhama Prakhya, Albert Lin, Aravinthan D.T. Samuel, Srinivas C. Turaga Cosyne 2021 We use stochastic threshold linear dynamics to model the C. elegans neural network, using connectome constraint on the neural activity. |
|
Lu Mi, Hao Wang, Yaron Meirovitch, Richard Schalek, Srinivas C. Turaga, Jeff W. Lichtman, Aravinthan D. T. Samuel, Nir Shavit MICCAI 2020 We show how to use deep learning to accelerate and optimize single-beam SEM acquisition of images. |
|
Yaron Meirovitch*, Lu Mi*, Hayk Saribekyan, Alexander Matveev, David Rolnick, Nir Shavit CVPR 2019 We introduce cross-classification clustering (3C), a technique that simultaneously tracks complex, interrelated objects in an image stack. |
|
Lu Mi*, Macheng Shen*, Jingzhao Zhang* We summarize our experiment results that compare these two categories of models GAN and VAE in terms of fidelity and mode collapse. |
|
|
|
|
|
The design and code of this website is adapted from Jon Barron's site. |