Research Scientist Deep Learning

Liang Zhao

https://scholar.google.com/citations?user=9xMR_iQAAAAJ&hl=en

*** ***** **** **** #*, Sunnyvale, CA 94085

Tel: 650-***-****

Email: *********@******.***.***

Professional Experience

2024- present

Mu Tech

AI Architect:

Working on GenAI, large language model, multi-modal learning for content(text, image, video, audio) understanding, generation(diffusion models), search, and recommendation. 2021-2024 Samsung Research America, CA

Staff II Machine Learning Engineer:

Developed deep learning models for image quality enhancement and investigated nerf- based 3D reconstruction.

Developed deep learning models for real time 3D avatar animation integrated to and commercialized as AR Emoji Apps on smart devices (AR/VR, smart phones, tablets, smart TVs).

Transferred Lip-sync technology to Visual Display team and Quacomm at HQ. Won A1 patent award 2023.

Won Star Award due to designing DL models well received by HQ and providing guide to team members in the area of expertise.

Won the third place in 2022 Innovation Challenge.

2016-2020 Baidu Research, Sunnyvale, CA

Sr. Research Scientist at Institute of Deep Learning: Developed deep learning algorithms/software for video analysis/generation, NLP, and news recommendation applications using unsupervised learning, reinforcement learning, and meta learning.

2013-2014 Intel, Santa Clara, CA

Software Engineer at Perceptual Computing RealSense SDK group. Developed object tracking using OpenCV and computer vision algorithms for the RealSense 3D camera SDK.

2013-2013 Perceptimed, Mountain View, CA

Sr. Research Scientist and Director:

Developed novel computer vision algorithms/software for pill identification. Developed novel anomaly detection algorithm for detecting segmentation errors. 2012-2013 Pearson KT, Menlo Park, CA

Sr. Research Scientist:

Developed algorithm/software for graphical/spatial data classification for automatic graph-based test scoring.

2010-2012 Indiana University, Bloomington, IN

Computer Scientist:

Developed dynamic programming algorithms for medical image analysis such as OCT image segmentation of nine retinal layers.

2007-2009 Digitalsmiths, Inc, Raleigh, NC

Computer Vision Scientist and R&D Manager:

‘ Developed algorithms and software for video analysis, face/object detection/ shape modeling, face recognition and tracking;

‘ Led a computer vision group on developing algorithms and software for image/ video indexing/retrieval, scene/texture classification. 2005-2006 Genex Technologies, Inc, Bethesda, MD

Computer Vision Staff Scientist and Team Lead:

‘ Developed algorithms and software for gesture recognition, object detection/ tracking/recognition/shape modeling, anomaly detection.

‘ Led a computer vision group on developing algorithms and software for navigation, image registration/modeling, structure from motion, stereo and motion correspondence, sensor fusion, 3D vision.

‘ Wrote proposals for seeking government funding.

2001-2005 University of Maryland, College Park, MD Research scientist:

‘ Developed computer vision algorithms and software for gesture recognition, object detection, tracking and recognition, event detection;

‘ Performed camera calibration and image quality evaluation.

‘ Wrote technical papers and funding proposals.

Computing Skills

Deep learning frameworks: PyTorch, Tensorflow, Caffe, PaddlePaddle(Baidu). Programming languages: C/C++, Python, Java, Matlab. Publications

Y. Li, J. Hestness, M. Elhoseiny, L. Zhao, K. Church, “Efficiently Disentangle Causal Representations,” First Conference on Parsimony and Learning (CPAL ), 2024. Y. Li, L. Zhao, K. Church, M. Elhoseiny, “Compositional Language Continual Learning,” ICLR 2020.

L. Zhao, Y. Wang, D. Dong, H. Tian, “Learning to Recommend via Meta Parameter Partition,” arXiv:1912.04108v1, 2019.

Y. Li, L. Zhao, J. Wang, J. Hestiness, “Compositional Generalization for Primitive Substitutions,” EMNLP, 2019.

L. Zhao, W. Xu, “Learning Good Representation via Continuous Attention,” arXiv:1903.12344v2, 2019.

T. Xu, A. Zhang, L. Zhao, “WALL-E: Efficient Reinforcement Learning Research Framework,” arXiv:1901.06086v2, 2019.

T. Xu, J. Peng, L. Zhao, Q. Liu, “Learning to Explore via Meta-Policy Gradient,” ICML 2018.

Y. Wang, Y. Yang, Z. Yang, L. Zhao, W. Xu, “Occlusion Aware Unsupervised Learning of Optical Flow,” CVPR 2018.

Z. Yang, P. Wang, W. Xu, L. Zhao, R. Nevatia, “Unsupervised Learning of Geometry with Edge-Aware Depth-Normal Consistency,” AAAI, 2018. L. Zhao, W. Wang, Y. Yang, W. Xu, “Unsupervised Learning Layers for Video Analysis,” arXiv:1705.08918, 2017.

L. Zhao, et al, “Semi-Automatic OCT Segmentation of Nine Retinal Layers”, ARVO

(The Association for Research in Vision and Ophthalmology) Annual Meeting, May 2012.

L. Zhao, L. Davis, “Closely Coupled Object Detection and Segmentation,” ICCV, Oct. 2005.

L. Zhao, L. Davis, “Segmentation and Appearance Model Building from An Image Sequence,” ICIP, Sep. 2005.

L. Zhao, L. Davis, “Iterative Figure-Ground Discrimination,” ICPR, Aug. 2004. L. Zhao, L. Davis, “Cooperative Body Part Labeling and Tracking,” University of Maryland, Technical Report, CS-TR-4587, 2004.

A. Mittal, L. Zhao, and L. Davis, “Human Body Pose Estimation Using Silhouette Shape Analysis,” IEEE Conf. on Advanced Video and Signal Based Surveillance, July, 2003.

Z. Yue, L. Zhao, and R. Chellappa, “View Synthesis of Articulating Humans Using Visual Hull,” Int'l Conf. on Multimedia and Expo, July, 2003. L. Zhao, G.S. Pingali, I. Carlbom, “Real-Time Head Orientation Estimation Using Neural Networks,” Int'l Conf. on Image Processing, Sept. 2002. L. Zhao, “Dressed human modeling, detection, and parts localization,” PhD Thesis, Carnegie Mellon University, Pittsburg, 2001.

L. Zhao, C. Thorpe, “Stereo- and Neural Network-based Pedestrian Detection,” IEEE Trans. on Intelligent Transportation Systems, Sep. 2000. L. Zhao, C. Thorpe, “Recursive Context Reasoning for Human Detection and Part Identification,” CVPR Workshop on Human Modeling, Analysis, and Synthesis, June 16, 2000.

L. Zhao, C. Thorpe, “Stereo and Neural Network-based Pedestrian Detection,” Proc. 1999 Int'l Conf. on Intelligent Transportation Systems, Tokyo, Japan, pp. 298-303, Oct. 5-7, 1999.

L. Zhao, G.S. Pingali, “Vision-Based Head Orientation Estimation in an Audio- Visual Tracking System,” Technical Memorandum 990812-02TM, Bell Labs, Lucent Technologies, Aug. 1999.

L. Zhao, C. Thorpe, “Qualitative and Quantitative Car Tracking from a Range Image Sequence,” CVPR, 1998.

L. Zhao, C. Thorpe, “Adaptive Vehicle Motion Estimation and Prediction,” Proc. 1998 SPIE Mobile Robots, Boston, Nov. 1-6, 1998.

L. Zhao, S.D. Ma, “Vision Based Robot Self Localization,” 12th Int'l Conf. on CAD/ CAM Robotics and Factories of the Future, London, 14-16 August, pp 139-144, 1996.

L. Zhao, “Applications of Fuzzy Mathematics in Computer Vision,” Master's Thesis, Chinese Academy of Sciences, Beijing, China, 1996. Patents

Spatial-Temporal Image Enhancement Using GNN-Transformer, filed in 2024.

Real-time Avatar Animation, filed in 2024.

Machine Learning Approach for Audio-Driven Avatar Animation or Other Functions, filed in 2023.

Light-Weight Machine Learning Models for Lip Sync Animation on Mobile Devices or Other Devices, filed in 2022.

Integrated systems and methods for video-based object modeling, recognition, and tracking

US Patent 8,170,280, 2012

Education

1996-2001 Carnegie Mellon University, Pittsburgh, PA, USA Ph.D. in Robotics, July 2001.

1993-1996 Chinese Academy of Sciences, Beijing, China M.S. in Pattern Recognition and Intelligent Control, June 1996. 1986-1991 Tsinghua University, Beijing, China

B.S. in Computer Science, July 1991.

Professional Activities

Paper Reviewer:

NeurIPS, ICML, ICLR, CVPR, Interspeech, ICCV, ECCV IEEE Trans. on Pattern Analysis and Machine Intelligence; Area Chair: ICLR, NeurIps, ICML

Won outstanding area chair at ICLR 2021.

Contact this candidate