Rebecca LI Data Scientist CV/ ML Engineer

E-mail: add5ej@r.postjobfree.com Tel: (1-832-***-**** EPD: Dec.2020

Locations: Willing to relocate GitHub Google Scholar Research Gate SUMMARY

Ph.D. Candidate with 5yrs research and industrial experience in Deep Learning, Machine Learning, and Computer Vision. The first author of the top-tier AI conference paper. Diverse advanced domain experience including Medical Image, Self-driving Car and Oil & Gas. Strong passion for real-world problems and high communication skills.

EDUCATION BACKGROUND

University of Houston, US (GPA: 3.9) 08/2015 – Now

• Ph.D. Candidate in Bio-image and Information Analysis Lab in Electrical and Computer Engineer Dept.

• Thesis: “Large Scale Nucleus Segmentation and Neighborhood Analysis on Rat Brain Images” XIDIAN University, China (GPA: 3.3) 09/2011 – 07/2015

• B.S in Information Engineering, School of Telecommunications Eng. PROFESSIONAL EXPERIENCE

University of Houston, Houston, TX Research Assistant, 08/2015 – Now Zero Human Effort Nucleus Cell Segmentation on Noisy Labels

• Created an iterative training framework based on supervised Mask RCNN to run in an unsupervised matter, especially when lacking high-quality training annotations in the bio-image field

• Solved the crowed-object challenges and false-negative detection errors from the pseudo annotations

• Achieved high fidelity segmentation: mean average precision 0.95, mean IoU=0.91 over 200,000 cells Co-location Spatial Pattern Analysis on Neuronal Neighborhood

• Selected significant neuron-neighbor pairs features by using hierarchical Tree LASSO

• Discovered the co-location and exclusive patterns by the association-rule based fast mining method

• Revealed the latent intercellular correlations, providing proofs for pathological and drug analysis Anadarko Petroleum, the Woodlands, TX Data Scientist Intern, 05 – 08/2019 Seismic Image Recovery and Optimal Sampling Recommendation

• Adapted a pixel inpainting neural network Wasserstein GAN to recover the compressed seismic images

• Achieved reliable recovery from compressed rates of 2 to 16; improved MSE 24% and 31% respectively, and performed 300 faster as compared with conventional approaches

• Proposed a non-uniform sampling strategy aiming for the optimal design of seismic signal acquisition

• Published in NeulPS Conference and the Leading Edge Journal Ambarella Corporation, San Jose, CA Deep Learning Engineer Intern, 05 – 08/2018 Cross-platform Solutions for Self-driving Car Chip Simulation

• Proposed cross-platform solutions over Keras, Pytorch, ONNX and the commercial python API for computer vision chips simulation, extending the options for user-specified platforms

• Achieved light hardware setting for instant segmentation and object detection algorithm testing

• Maintained IOU to 97% ~ 99% from the coarsest to the finest chips’ quantization setting NINDS, National Institute of Health, Bethesda, MD Pre-doc Fellow, 05 – 08/2017 A Pipeline for Image Processing and Data Analysis of Brain Tissues LinkedIn

• Accelerated large-scale image alignment by 10 with uniform keypoint control and multiprocessing

• Reconstructed the whole cell bodies including cell detection, soma extraction, and processes tracing

• Extended format conversions to third-party cytometry analysis software for real-time cell visualization RELATED PROJECTS

• [DA] Predicting the House Price from 1950-2010 Report 11/2018

• [CV] Multiplex Image Denoising by Wavelet Transform Report 05/2018

• [CV] Pixel Translator from Sketch to Colored Image: An Application of GAN Report 10/2017

• [AI] Solving Pick-up and Drop-off (PD) Problem of Reinforcement Learning Report 10/ 2016

• [ML] In-depth Comparisons on Feature Reduction & Classification of High-dim Images Report 02/2016 TECHNICAL SKILLS

• Programming Skills: Python, Matlab, R, Scikit-image, OpenCV, Pandas, C, C++, VHDL

• Frameworks and Platforms: Linux, Keras, Tensorflow, Pytorch, Caffe, ONNX, Docker, FCS Express LANGUAGES language

• Languages: Mandarin, English

LEADERSHIP EXPERIENCE

• Graduate Affair Chair, IEEE-University of Houston Student Branch 2018

• Outreach/ Coordinator, US & China Innovation and Investment Summit 2017

• President, Model United Nations Association of Xidian University 2013 AWARDS

• Winning Poster, Mission Connect /TIRR Foundation Annual Scientific Symposium 2017

• Presidential Scholarship, University of Houston 2015 REVIEWERS

• Geophysical Journal International 2020

• IEEE International Symposium on Biomedical Imaging 2018 PUBLICATIONS

• Li, X.R., Mitsakos, N., Lu, P., Xiao, Y., Zhan, C. and Zhao, X., 2019. Generative Inpainting Network Applications on Seismic Image Compression and Non-Uniform Sampling. NeurIPS, Deep Inverse Workshop (2019).

• Li, Xiaoyang Rebecca, et al. "Seismic Compressive Sensing by Generative Inpainting Network: Toward an Optimized Acquisition Survey." The Leading Edge 38.12 (2019): 923-933.

• Yuan, P., Rezvan, A., Li, X., Varadarajan, N. and Van Nguyen, H., 2019. Phasetime: Deep Learning Approach to Detect Nuclei in Time Lapse Phase Images. Journal of clinical medicine, 8(8), p.1159.

• Yuan,P., Mobiny,A., Jahanipour,J. Li,X., et al. "Few Is Enough: Task-Augmented Active Meta-Learning for Brain Cell Classification", MICCAI (2020) (Accepted)

• Zhao X, Lu P, Zhang Y, Chen J, Li X. Swell-noise attenuation: A deep learning approach. The Leading Edge. 2019 Dec;38(12):934-42.

• Li Xiaoyang, “A Simplified Normalization Operation for Perfect Reconstruction from a Modified STFT”, In Pros, IEEE 12th International Conference on Signal Processing (ICSP), 2014, P42-45

Contact this candidate