Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Chun Pong LauSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works34 from public data
- DiffProtect: Generative adversarial examples using diffusion models for facial privacy protection61
This work proposes DiffProtect, which utilizes a diffusion autoencoder to generate semantically meaningful perturbations on FR systems, and demonstrates that DiffProtect produces more natural-looking encrypted images than state-of-the-art methods while achieving significantly higher attack success rates.
- Threats and Defenses in the Federated Learning Life Cycle: A Comprehensive Survey and Challenges55
This article reviews the most representative and state-of-the-art threats and defense frameworks throughout the FL service life cycle, and dives into the defense frameworks, analyzing the relationship between threats and defenses, and compare the trade-offs among different defense strategies.
- Machine learning methods for protein-protein binding affinity prediction in protein design38
The prediction methods and associated datasets are reviewed and the requirements and construction methods of binding affinity prediction models for protein design are discussed, revealing the potential of a paradigm shift in protein design.
- Semi-Supervised Crowd Counting With Contextual Modeling: Facilitating Holistic Understanding of Crowd Scenes33
This paper presents a new semi-supervised method based on the mean teacher framework that achieves the state-of-the-art performance, surpassing previous approaches by a large margin on challenging benchmarks such as ShanghaiTech A and UCF-QNRF.
- A Gray-Box Attack Against Latent Diffusion Model-Based Image Editing by Posterior Collapse29
This work proposes the Posterior Collapse Attack (PCA), a novel framework for protecting images from unauthorized manipulation that outperforms existing techniques in protection effectiveness, computational efficiency (runtime and VRAM), and generalization across VAE-based LDM variants.
- T2ICount: Enhancing Cross-Modal Understanding for Zero-Shot Counting26
A Hierarchical Semantic Correction Module that progressively refines text-image feature alignment, and a Representational Regional Coherence Loss that provides reliable supervision signals by leveraging the cross-attention maps extracted from the demising U-Net are proposed.
- Instant Adversarial Purification with Adversarial Consistency Distillation26
This paper introduces One Step Control Purification (OSCP), a novel defense framework that achieves robust adversarial purification in a single Neural Function Evaluation (NFE) within diffusion models, and proposes Gaussian Adversarial Noise Distillation (GAND) as the distillation objective and Controlled Adversarial Purification (CAP) as the inference pipeline.
- Artwork Protection Against Neural Style Transfer Using Locally Adaptive Adversarial Color Attack23
Locally Adaptive Adversarial Color Attack (LAACA) is proposed, empowering artists to protect their artwork from unauthorized style transfer by processing before public release, and the Adversarial Color Distance Metric (ACDM), designed to quantify the color difference of images pre- and post-manipulations.
- 20
- A White-Box False Positive Adversarial Attack Method on Contrastive Loss Based Offline Handwritten Signature Verification Models20
A novel false positive attack method is proposed that treats the attack as a style transfer between closely related but distinct writing styles, and two new loss functions are introduced that enhance the attack success rate and guide the generation of deceptive images.
- 17
- 15
- A Siamese Transformer Network for Zero-Shot Ancient Coin Classification15
The Double Siamese ViT model is shown to surpass the state of the art by a large margin, achieving an overall accuracy of 81% and the majority of the method’s errors are unrelated to the intrinsic aspects of the algorithm itself, but are rather a consequence of unclean data.
- Achieving fair medical image segmentation in foundation models with adversarial visual prompt tuning14
- 14
- 10
- Beyond Vulnerabilities: A Survey of Adversarial Attacks as Both Threats and Defenses in Computer Vision Systems8
A systematic analysis of adversarial attack methodologies across three primary domains: pixel-space attacks, physically realizable attacks, and latent-space attacks, which reveals critical research gaps, particularly in neural style transfer protection and computational efficiency requirements.
- My Face Is Mine, Not Yours: Facial Protection Against Diffusion Model Face Swapping6
This paper introduces a novel proactive defense strategy through adversarial attacks that preemptively protect facial images from being exploited by diffusion-based deepfake systems.
- GaitProtector: Impersonation-Driven Gait De-Identification via Training-Free Diffusion Latent Optimization4
This work proposes GaitProtector, an impersonation-driven gait deidentification framework that formulates privacy protection as a unified objective with two tightly coupled components: obfuscation and impersonation, which repels the protected gait from the source identity and attracts it toward a selected target identity.
- 2
- 2
- 2
- 2
- GazeSync: A Mobile Eye-Tracking Tool for Analyzing Visual Attention on Dynamically Manipulated Content1
GazeSync is presented, a reusable mobile system that synchronizes on-device gaze estimation with real-time image transformation matrices (scale, rotation, and translation) and enables the accurate reconstruction of image-relative attention patterns, decoupling visual attention from device interaction.
- 1
- 1
- Noise-Aware and Dynamically Adaptive Federated Defense Framework for SAR Image Target Recognition1
NADAFD is proposed, a noise-aware and dynamically adaptive federated defense framework that integrates frequency-domain, spatial-domain, and client-behavior analyses to counter SAR-specific backdoor threats and introduces a frequency-domain collaborative inversion mechanism to expose cross-client spectral inconsistencies indicative of hidden backdoor triggers.
- 1
- A U-net and KMeans based method for brain tumor segmentation and measurement1
The U-net algorithm is improved by adding ResBlock and Batch Normalization modules, as well as increasing the network depth, so as to improve its accuracy for tumor site segmentation in medical images of the brain.
- –
- –
- –
- RVINN: a flexible modeling for inferring dynamic transcriptional and post-transcriptional regulation using physics-informed neural networks–
This model has the ability to computationally revalidate and visualize dynamic biological phenomena, such as transcriptional ripple, co-bursting, and buffering in a breast cancer cell line, and suggests putative molecular mechanisms underlying these phenomena.
- Breaking PEFT Limitations: Leveraging Weak-to-Strong Knowledge Transfer for Backdoor Attacks in LLMs–
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.