New GENADA framework generates efficient adversarial attacks for time series models

Researchers have developed GENADA, a novel generative adversarial attack framework designed to create deceptive perturbations for time series deep learning models. This framework learns a generative model to produce these perturbations efficiently in a single forward pass, offering a computationally less burdensome alternative to traditional gradient-based attacks. GENADA has demonstrated comparable attack quality to existing methods while significantly reducing the time required for perturbation generation during inference. AI

IMPACT This research could lead to more robust defenses against adversarial attacks in critical time series applications.

RANK_REASON The cluster contains a research paper detailing a new framework for adversarial attacks on time series models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• GENADA
• Gotit.pub
• Hugging Face
• IArxiv
• ScienceCast

• paper
• other

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