In reversible image secret sharing (RISS), the cover image can be recovered to some degree, and a share can be comprehensible rather than noise-like. Reversible cover images play an important role in law enforcement and medical diagnosis. The comprehensible share can not only reduce the suspicion of attackers but also improve the management efficiency of shares. In this paper, we first provide a formal definition of RISS. Then, we propose an RISS algorithm for a $(k,n)$ -threshold based on the principle of the Chinese remainder theorem-based ISS (CRTISS). In the proposed RISS, the secret image is losslessly decoded by a modular operation, and the original cover image is recovered by a binarization operation, both of which are just simple operations. Theoretical analyses and experiments are provided to validate the proposed definition and algorithm.

Malicious traffic classification is crucial for Intrusion Detection Systems (IDS). However, traditional Machine Learning approaches necessitate expert knowledge and a significant amount of well-labeled data. Although recent studies have employed pre-training models from the Natural Language Processing domain, such as ET-BERT, for traffic classification, their effectiveness is impeded by limited input length and fixed Byte Pair Encoding. To address these challenges, this paper presents Flow-MAE, a pre-training model that employs Masked AutoEncoders (MAE) from the Computer Vision domain to achieve accurate, efficient, and robust malicious network traffic classification. Flow-MAE overcomes these challenges by utilizing burst (a generic representation of network traffic) and patch embedding to accommodate extensive traffic length. Moreover, Flow-MAE introduces a self-supervised pre-training task, the Masked Patch Model, which captures unbiased representations from bursts with varying lengths and patterns. Experimental results from six datasets reveal that Flow-MAE achieves new state-of-the-art accuracy (>0.99), efficiency (>900 samples/s), and robustness across diverse network traffic types. In comparison to the state-of-the-art ET-BERT, Flow-MAE exhibits improvements in accuracy and speed by 0.41%-1.93% and 7.8x-10.3x, respectively, while necessitating only 0.2% FLOPs and 44% memory overhead. The efficacy of the core designs is validated through few-shot learning and ablation experiments. The code is publicly available at https://github.com/NLear/Flow-MAE.

Thien-and-Lin’s polynomial-based secret image sharing (PSIS) is utilized as the basic method to achieve PSISs with better performances, such as meaningful shares, two-in-one property and shares with different priorities. However, this $(k,n)$ threshold PSIS cannot achieve lossless recovery for pixel values more than 250. Furthermore, current solutions to lossless recovery for PSIS have several natural drawbacks, such as large computational costs and random pixel expansion. In this paper, a lossless and efficient $(k,n)$ threshold PSIS scheme with reduced shadow size is presented. For lossless recovery and efficiency, two adjacent pixels are specified as a secret value, the prime in the sharing polynomial is replaced with 65,537, and then the additional screening operation can ensure each shared value in the range $[0,\mathrm{65,535}]$. To reduce shadows size and improve security, only the first $k-1$ coefficients are embedded with secret values and the last coefficient is assigned randomly. To prevent the leakage of secrets, generalized Arnold permutation with special key generating strategy is performed on the secret image prior to sharing process without key distribution. Both theoretical analyses and experiments are conducted to demonstrate the effectiveness of the proposed scheme.