Search Results for author: Siqi Ma

Found 12 papers, 1 papers with code

Parameter-Saving Adversarial Training: Reinforcing Multi-Perturbation Robustness via Hypernetworks

no code implementations28 Sep 2023 Huihui Gong, Minjing Dong, Siqi Ma, Seyit Camtepe, Surya Nepal, Chang Xu

Adversarial training serves as one of the most popular and effective methods to defend against adversarial perturbations.

Stealthy Physical Masked Face Recognition Attack via Adversarial Style Optimization

no code implementations18 Sep 2023 Huihui Gong, Minjing Dong, Siqi Ma, Seyit Camtepe, Surya Nepal, Chang Xu

Moreover, to ameliorate the phenomenon of sub-optimization with one fixed style, we propose to discover the optimal style given a target through style optimization in a continuous relaxation manner.

Face Recognition

Reducing Adversarial Training Cost with Gradient Approximation

no code implementations18 Sep 2023 Huihui Gong, Shuo Yang, Siqi Ma, Seyit Camtepe, Surya Nepal, Chang Xu

Deep learning models have achieved state-of-the-art performances in various domains, while they are vulnerable to the inputs with well-crafted but small perturbations, which are named after adversarial examples (AEs).

TransFlow: Transformer as Flow Learner

no code implementations CVPR 2023 Yawen Lu, Qifan Wang, Siqi Ma, Tong Geng, Yingjie Victor Chen, Huaijin Chen, Dongfang Liu

Optical flow is an indispensable building block for various important computer vision tasks, including motion estimation, object tracking, and disparity measurement.

Motion Estimation object-detection +4

Two-in-one Knowledge Distillation for Efficient Facial Forgery Detection

no code implementations21 Feb 2023 Chuyang Zhou, Jiajun Huang, Daochang Liu, Chengbin Du, Siqi Ma, Surya Nepal, Chang Xu

More specifically, knowledge distillation on both the spatial and frequency branches has degraded performance than distillation only on the spatial branch.

Knowledge Distillation Vocal Bursts Valence Prediction

Anti-Compression Contrastive Facial Forgery Detection

no code implementations13 Feb 2023 Jiajun Huang, Xinqi Zhu, Chengbin Du, Siqi Ma, Surya Nepal, Chang Xu

To enhance the performance for such models, we consider the weak compressed and strong compressed data as two views of the original data and they should have similar representation and relationships with other samples.

Contrastive Learning

Private Image Generation With Dual-Purpose Auxiliary Classifier

no code implementations CVPR 2023 Chen Chen, Daochang Liu, Siqi Ma, Surya Nepal, Chang Xu

However, apart from this standard utility, we identify the "reversed utility" as another crucial aspect, which computes the accuracy on generated data of a classifier trained using real data, dubbed as real2gen accuracy (r2g%).

Image Generation Privacy Preserving

Solve the Puzzle of Instance Segmentation in Videos: A Weakly Supervised Framework with Spatio-Temporal Collaboration

no code implementations15 Dec 2022 Liqi Yan, Qifan Wang, Siqi Ma, Jingang Wang, Changbin Yu

Instance segmentation in videos, which aims to segment and track multiple objects in video frames, has garnered a flurry of research attention in recent years.

Depth Estimation Instance Segmentation +2

DeepFake Disrupter: The Detector of DeepFake Is My Friend

no code implementations CVPR 2022 Xueyu Wang, Jiajun Huang, Siqi Ma, Surya Nepal, Chang Xu

We argue that the detectors do not share a similar perspective as human eyes, which might still be spoofed by the disrupted data.

Face Swapping

Wideband Channel Estimation for IRS-Aided Systems in the Face of Beam Squint

no code implementations5 Jun 2021 Siqi Ma, Wenqian Shen, Jianping An, Lajos Hanzo

Intelligent reflecting surfaces (IRSs) improve both the bandwidth and energy efficiency of wideband communication systems by using low-cost passive elements for reflecting the impinging signals with adjustable phase shifts.

Backdoor Attacks and Countermeasures on Deep Learning: A Comprehensive Review

1 code implementation21 Jul 2020 Yansong Gao, Bao Gia Doan, Zhi Zhang, Siqi Ma, Jiliang Zhang, Anmin Fu, Surya Nepal, Hyoungshick Kim

We have also reviewed the flip side of backdoor attacks, which are explored for i) protecting intellectual property of deep learning models, ii) acting as a honeypot to catch adversarial example attacks, and iii) verifying data deletion requested by the data contributor. Overall, the research on defense is far behind the attack, and there is no single defense that can prevent all types of backdoor attacks.

Revocable Federated Learning: A Benchmark of Federated Forest

no code implementations8 Nov 2019 Yang Liu, Zhuo Ma, Ximeng Liu, Zhuzhu Wang, Siqi Ma, Ken Ren

A learning federation is composed of multiple participants who use the federated learning technique to collaboratively train a machine learning model without directly revealing the local data.

Federated Learning

Cannot find the paper you are looking for? You can Submit a new open access paper.