Federated learning is a novel distributed learning framework, which enables thousands of participants to collaboratively construct a deep learning model. In order to protect confidentiality of the training data, the shared information between server and participants are only limited to model parameters. However, this setting is vulnerable to model poisoning attack, since the participants have permission to modify the model parameters. In this paper, we perform systematic investigation for such threats in federated learning and propose a novel optimization-based model poisoning attack. Different from existing methods, we primarily focus on the effectiveness, persistence and stealth of attacks. Numerical experiments demonstrate that the proposed method can not only achieve high attack success rate, but it is also stealthy enough to bypass two existing defense methods.

No abstract

No abstract

No abstract

Modern adaptive image steganography with minimizing a distortion function has high performance of undetectability. However, when an image with hidden information is attacked by JPEG compression, its robustness cannot be guaranteed, that remarkably limits its extension from the lab to the real world. In this paper, a novel image steganographic algorithm is proposed that is robust to JPEG compression. First, by using the sign of DCT coefficients, that remains unchangeable before and after JPEG compression, we select the candidate coefficients for resisting JPEG compression. Second, the designed distortion function assigns cost for each candidate DCT coefficient. Finally, relying on both error correction code and Syndrome-Trellis Codes, an encoded message is embedded into the cover image with minimum embedding distortion. Compared with prior arts, extensive experimental results highlight both undetectability and robustness of our proposed algorithm. INDEX TERMS Robust steganography, distortion function, JPEG compression, sign of DCT coefficients.