Talk on "Privacy-Preserving Generative Model for Images"
Title of the talk:
Privacy-Preserving Generative Model for Images
Prof. Chun-Shien Lu, Institute of Information Science, Academia Sinica
16:00-17:30 JST / Wednesday, August 17th, 2022
If you would like to join, please contact by email.
Email : echizen-sec [at] nii.ac.jp
The person in charge will send you the zoom url of this event.
In this talk, I will introduce our works, published in CVPR 2021 and 2022, on Privacy-Preserving Generative Model for Images.
With the growing use of camera devices, the industry has many image datasets that provide more opportunities for collaboration between the machine learning community and industry. However, the sensitive information in the datasets discourages data owners from releasing these datasets. Despite recent research devoted to removing sensitive information from images, they provide neither meaningful privacy-utility trade-off nor provable privacy guarantees. In the first work, with the consideration of the perceptual similarity, we propose perceptual indistinguishability (PI) as a formal privacy notion particularly for images. We also propose PI-Net, a privacy-preserving mechanism that achieves image obfuscation with PI guarantee. Our study shows that PI-Net achieves significantly better privacy utility trade-off through public image data.
On the other hand, one may use differentially private generative models to generate synthetic data. Unfortunately, generators are typically restricted to generating images of low-resolutions due to the limitation of noisy gradients. In the second work, we propose DPGEN, a network model designed to synthesize high-resolution natural images while satisfying differential privacy. In particular, we propose an energy-guided network trained on sanitized data to indicate the direction of the true data distribution via Langevin Markov chain Monte Carlo (MCMC) sampling method. In contrast to the state-of-the-art methods that can process only low-resolution images (e.g., MNIST and Fashion-MNIST), DPGEN can generate differentially private synthetic images with resolutions up to 128X128 with superior visual quality and data utility.
Bio of the speaker:
Chun-Shien Lu is currently a full research fellow in the Institute of Information Science, Academia Sinica, Taipei, Taiwan. His current research interests mainly focus on algorithms and applications of compressed sensing, various topics (including signal processing and security) of multimedia, and deep learning.
Dr. Lu serves as Area Chairs of ICASSP 2012~2014, ICIP 2013, ICME 2018, ICIP 2019-2022, ICML 2020, ICLR 2021~2022, NeurIPS 2022, and ACM Multimedia 2022. Dr. Lu has owned four US patents, five ROC patents, and one Canadian patent in digital watermarking and graphic QR code. Dr. Lu won Ta-You Wu Memorial Award, National Science Council (Taiwan) in 2007 and was a co-recipient of a National Invention and Creation Award in 2004.
Dr. Lu was an associate editor of IEEE Trans. on Image Processing (2010/12~2014) and serves as the 2nd term of associate editor of IEEE Trans. on Image Processing since March 2018.