Home Knowledge Base FADRW

FADRW

No mentions found

This entity hasn't been tracked yet, or Iris is still building its knowledge base.

Related Articles from SNS

FADRW: A Feature-Aware Modulated and Dynamically Reweighted Loss for Few-Shot Linguistic Steganalysis

Announce Type: cross Abstract: The ubiquity of social media platforms facilitates malicious linguistic steganography, posing significant security risks. However, detection is severely hampered by two fundamental issues during model training. Firstly, extreme class imbalance (less than 1% steganographic samples) induces a strong decision bias.

arXiv CS 1d ago