FedOT framework enhances ownership verification and leakage tracing for federated LDMs

Researchers have introduced FedOT, a new framework designed to verify ownership and trace leakage in federated latent diffusion models (LDMs). This system addresses vulnerabilities in existing methods by incorporating a chunked watermarking approach for both ownership verification and client identification. Additionally, FedOT employs Latent Vector Transformation (LVT) to prevent watermark removal attacks by degrading image quality if the VAE is replaced, ensuring the model's integrity and traceability. AI

IMPACT This framework could improve security and trust in federated learning applications involving generative models.

RANK_REASON The cluster contains a research paper detailing a new framework for federated latent diffusion models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Hugging Face Daily Papers →

• Federated Learning
• Latent Diffusion Models
• Latent Vector Transformation

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →