Publications

Can Textual Unlearning Solve Cross-Modality Safety Alignment?
Trishna Chakraborty, Erfan Shayegani, Zikui Cai, Nael B. Abu-Ghazaleh, M. Salman Asif, Yue Dong, Amit Roy-Chowdhury, Chengyu Song
EMNLP 2024
Disguise without Disruption: Utility-Preserving Face De-Identification
Zikui Cai, Zhongpai Gao, Benjamin Planche, Meng Zheng, Terrence Chen, M. Salman Asif, Ziyan Wu
We developed an algorithm that seamlessly de-identifies facial images without compromising utility.
AAAI 2024
Ensemble-based Blackbox Attacks on Dense Prediction
Zikui Cai*, Yaoteng Tan*, M. Salman Asif
We perform query efficient blackbox attacks on dense prediction models of different tasks simultaneously through weight balancing and weight optimization on surrogates.
CVPR 2023
Blackbox Attacks via Surrogate Ensemble Search
Zikui Cai, Chengyu Song, Srikanth V. Krishnamurthy, Amit K. Roy-Chowdhury, M. Salman Asif
We generate extremely query efficient blackbox attacks via searching in surrogate ensemble weight space.
NeurIPS 2022
Zero-Query Transfer Attacks on Context-Aware Object Detectors
Zikui Cai, Shantanu Rane, Alejandro E. Brito, Chengyu Song, Srikanth V. Krishnamurthy, Amit K. Roy-Chowdhury, M. Salman Asif
We attack context-aware black-box object detectors in a zero-query setting, by leveraging context information and estimated success probability.
CVPR 2022
arXiv / open access / BIB
@inproceedings{cai2022zero, title={Zero-Query Transfer Attacks on Context-Aware Object Detectors}, author={Cai, Zikui and Rane, Shantanu and Brito, Alejandro E and Song, Chengyu and Krishnamurthy, Srikanth V and Roy-Chowdhury, Amit K and Asif, M Salman}, booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition}, pages={15024--15034}, year={2022} }
Context-Aware Transfer Attacks for Object Detection
Zikui Cai, Xinxin Xie, Shasha Li, Mingjun Yin, Chengyu Song, Srikanth V. Krishnamurthy, Amit K. Roy-Chowdhury, M. Salman Asif
We transfer attack black-box object detectors by leveraging context information, including co-occurrence of objects and their relative locations and sizes.
AAAI 2022
arXiv / code / slides / poster / talk / BIB
@inproceedings{cai2022context, title={Context-aware transfer attacks for object detection}, author={Cai, Zikui and Xie, Xinxin and Li, Shasha and Yin, Mingjun and Song, Chengyu and Krishnamurthy, Srikanth V and Roy-Chowdhury, Amit K and Asif, M Salman}, booktitle={Proceedings of the AAAI Conference on Artificial Intelligence}, volume={36}, number={1}, pages={149--157}, year={2022} }
Exploiting Multi-Object Relationships for Detecting Adversarial Attacks in Complex Scenes
Mingjun Yin, Shasha Li, Zikui Cai, Chengyu Song, M. Salman Asif, Amit K. Roy-Chowdhury, Srikanth V. Krishnamurthy
We developed a novel approach (SCENE-Lang) to perform model-agnostic context consistency checks using language models.
ICCV 2021
arXiv / open access / BIB
@inproceedings{yin2021exploiting, title={Exploiting multi-object relationships for detecting adversarial attacks in complex scenes}, author={Yin, Mingjun and Li, Shasha and Cai, Zikui and Song, Chengyu and Asif, M Salman and Roy-Chowdhury, Amit K and Krishnamurthy, Srikanth V}, booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision}, pages={7858--7867}, year={2021} }
Data-Driven Illumination Patterns For Coded Diffraction Imaging
Zikui Cai, Rakib Hyder, M. Salman Asif
We present a framework to optimize the sensing parameters to recover signals from coded diffraction patterns using alternating minimization-based phase retrieval method.
ICIP 2021
paper / BIB
@inproceedings{cai2021data, title={Data-driven illumination patterns for coded diffraction imaging}, author={Cai, Zikui and Hyder, Rakib and Asif, M Salman}, booktitle={2021 IEEE International Conference on Image Processing (ICIP)}, pages={2818--2822}, year={2021}, organization={IEEE} }
Solving Phase Retrieval with a Learned Reference
Rakib Hyder*, Zikui Cai*, M. Salman Asif
(* co-first authorship)
We solve phase retrieval via unrolled network and learned reference that provide near-perfect recovery at fixed and small computational cost.
ECCV 2020
arXiv / code / BIB
@inproceedings{hyder2020solving, title={Solving phase retrieval with a learned reference}, author={Hyder, Rakib and Cai, Zikui and Asif, M Salman}, booktitle={European Conference on Computer Vision}, pages={425--441}, year={2020}, organization={Springer} }