Welcome

About

I am now a Research Associate at the University of New South Wales (UNSW). I used to be a Senior Algorithm Engineer at DAMO Academy of Alibaba Group.

My major research focuses on Efficient Deep Learning in Neural Architecture Search for Computer Vision, with their applications in Classification, Object Detection, 3D Action Recognition, Deployments on IoT Devices and other real-world tasks.

The second research focus is Learned Image/Video Compression, to design the next generation Codec for multimedia.

My google scholar site: https://scholar.google.com/citations?user=eDiXHP8AAAAJ

Projects

Lightweight Neural Architecture Search (Light-NAS) Main contributor 2021 -- Now

Light-NAS is an integrated, distributed, full-stack framework for Training-free Neural Architecture Search based on Pytorch and OpenMPI. It is able to design efficient deep neural networks for Classfication, Detection, 3D Action Recognition within 1 GPU day. It also includes Latency Prediction Search for specific hardware, such as a GPU, mobile and IoT Devices.

Entropy-Driven Mixed-Precision Quantization for Deep Network Design, Accepted by NeurIPS 2022.
MAE-DET: Revisiting maximum entropy principle in zero-shot NAS for efficient object detection, Accepted by ICML 2022.
In actual applications, speed up Face Detection model 3x, Reid model 1.68x, Car Plate Recognition model 2.5x.

Github: https://github.com/alibaba/lightweight-neural-architecture-search

Learned Image/Video Compression Main contributor 2018 -- 2021

Learned compression methods were proposed to use an entropy model to approximate the distribution of the compressible latents with CNNs, showing promising performance comparable to traditional image codecs. We propose Interpolation Variable Rate model and Spatiotemporal Entropy for Image/Video Compression.

4 Tracks Winner of the Challenge on Learned Image Compression on CVPR 2019.
Interpolation variable rate image compression, Accepted by ACM MM 2021.

Github: https://github.com/tinyvision/IPCodec

Employment History

Research Associate SEIT and CSE UNSW 2022.08 – Now
Algorithm Engineer Damo Academy Alibaba Group 2020.03 – 2022.07
Algorithm Engineer Research Institute Tuya Co., Ltd (Startup) 2018.08 – 2020.03
Hardware Engineer 2012 Laboratory Huawei Group 2017.07 – 2018.08

Education

Master's Degree Optical Engineering Nanjing University 2014 – 2017
Bachelor's Degree Information Engineering Nanjing University 2010 – 2014

Representative Research Publications

Sun, Z., Ge, C., Wang, J., Lin, M., Chen, H., Li, H., & Sun, X. (2022). Entropy-driven mixed-precision quantization for deep network design on iot devices. Thirty-sixth Conference on Neural Information Processing Systems (NeurIPS).
Sun, Z., Lin, M., Sun, X., Tan, Z., Li, H., & Jin, R. (2022). Mae-det: Revisiting maximum entropy principle in zero-shot nas for efficient object detection. International Conference on Machine Learning (ICML).
Li, D., Sun, Z., Tan, Z., Xiuyu Sun, F. Z., Qian, Y., & Li, H. (2022). Jmpnet: Joint motion prediction for learning-based video compression. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 1855–1859.
Sun, Z., Tan, Z., Sun, X., Zhang, F., Qian, Y., Li, D., & Li, H. (2021). Interpolation variable rate image compression. ACM International Conference on Multimedia (ACM MM), 5574–5582.
Lin, M., Wang, P., Sun, Z., Chen, H., Sun, X., Qian, Q., Li, H., & Jin, R. (2021). Zen-nas: A zero-shot nas for high-performance deep image recognition. IEEE/CVF International Conference on Computer Vision (ICCV).
Zhou, L., Sun, Z., Wu, X., & Wu, J. (2019). End-to-end optimized image compression with attention mechanism. 4 Tracks Winner Invitation Paper on CVPR.
Zhang, X., Sun, Z., Shan, Y., Li, Y., Wang, F., Zeng, J., & Zhang, Y. (2017). A high performance distributed optical fiber sensor based on ϕ-otdr for dynamic strain measurement. IEEE Photonics Journal, 9(3), 1–12. https://doi.org/10.1109/JPHOT.2017.2700020

Page updated

Google Sites

Report abuse