Existing linguistic steganography methods primarily rely on content transformations to conceal secret messages. However, they often cause subtle yet looking-innocent deviations between normal and stego texts, posing potential security risks in real-world applications. To address this challenge, we propose a content-preserving linguistic steganography paradigm for perfectly secure covert communication without modifying the cover text. Based on this paradigm, we introduce CLstega (Content-preserving Linguistic steganography), a novel method that embeds secret messages through controllable distribution transformation. CLstega first applies an augmented masking strategy to locate and mask embedding positions, where MLM (masked language model)-predicted probability distributions are easily adjustable for transformation. Subsequently, a dynamic distribution steganographic coding strategy is designed to encode secret messages by deriving target distributions from the original probability distributions. To achieve this transformation, CLstega elaborately selects target words for embedding positions as labels to construct a masked sentence dataset, which is used to fine-tune the original MLM, producing a target MLM capable of directly extracting secret messages from the cover text. This approach ensures perfect security of secret messages while fully preserving the integrity of the original cover text. Experimental results demonstrate that CLstega can achieve a 100% extraction success rate, and outperforms existing methods in security, effectively balancing embedding capacity and security.

Existing modification-based linguistic steganography methods primarily perform linguistic manipulations within a single embedding space to conceal secret information. However, these methods are stringently constrained by the original semantics of the cover text, making it struggle to achieve a satisfactory embedding capacity in a single embedding space. In this paper, we propose a novel Multi-granularity Modification-based Linguistic Steganography framework (MMLS) that hides secret information in both syntactic space and symbolic space, enhancing syntactic naturalness and semantic coherence while further increasing embedding capacity. Specifically, MMLS utilizes a paraphrase generation model to automatically modify the syntactic structure of the given original sentence, which enables the generation of paraphrases and the preservation of semantics simultaneously. Moreover, MMLS employs a distance-aware syntactic bins coding strategy to embed part of secret information into the syntactic space. This strategy utilizes a cluster-based way to partition the implicit syntactic space into a finite number of separate zones, thus increasing the number of candidate paraphrases and avoiding the selection of semantically distorted steganographic texts. Finally, the pre-trained BERT is used to replace some words in candidate paraphrases with their synonyms. Such a design embeds the remaining secret information into symbolic space while ensuring syntactic and semantic naturalness. Experimental results demonstrate that MMLS significantly outperforms existing methods in terms of semantic coherence, embedding capacity, and security.

Traditional backdoor watermarking techniques typically rely on injecting trigger samples into the training process, which can lead to model overfitting, distort feature distributions, and compromise decision boundaries, ultimately degrading the performance of the primary task. To address these limitations, we propose DualPrivMark, a dual-layer backdoor watermarking scheme that leverages private model embedding. Instead of directly embedding watermark into the protected model, Dual-PrivMark constructs a dedicated private model that cooperates with the target model, thereby isolating the watermarking task from the original task and preserving task fidelity. Within the private model, we design a two-layer watermarking architecture that combines a label layer for trigger-based verification with a signature layer for image-based authentication, enabling both high-fidelity embedding and enhanced robustness. Furthermore, a multi-task learning strategy is employed to jointly optimize the original and watermarking tasks, ensuring high accuracy across both. Experimental results demonstrate that, compared with existing methods, DualPrivMark achieves reliable watermark embedding while substantially improving model fidelity, and providing strong resistance against ambiguity attacks.

To enhance the embedding capacity, the existing linguistic steganography methods predominantly focus on the word or phrase level, with limited emphasis on the sentence level. Nevertheless, these approaches exhibit a deficiency in achieving an optimal balance between embedding capacity and semantic coherence. Moreover, compromised semantic coherence can potentially increase security risks. In this paper, we propose a novel sentence-levelSteganography framework toHideInformation inSyntaxSpace (HISS-Stega) that enables larger embedding capacity while preserving better semantic coherence. Specifically, HISS-Stega builds a syntax-controlled paraphrase generation model to automatically modify the expression forms of the covertext, thereby augmenting the diversity of transformations. This enhancement contributes to the overall improvement in embedding capacity. Subsequently, a syntactic bins coding strategy is employed for successfully embedding secret information in the generated syntax space. Furthermore, HISS-Stega incorporates a semantic distortion function aimed at identifying the optimal syntactic structure for concealing secret information, thereby ensuring enhanced semantic coherence and mitigating potential security risks. The experimental results demonstrate that, in comparison to existing methods, HISS-Stega not only enhances the embedding capacity of each sentence but also maintains a high level of semantic coherence and anti-steganalysis capability.

The autoregressive-based model with the left-to-right generation order has been a predominant paradigm for generative linguistic steganography. However, such steganography does not perform well on semantic control and content planning, which is forced by the secret message during the generation process. To mitigate this issue and efficiently produce high-quality steganographic texts (stegotexts), we present aProgressiveNon-autoregressiveGenerative linguisticSteganography (PNG-Stega), which encodes secret messages and extends the context to generate stegotexts in a multi-round insertion manner. Each round continuously refines the generated steganographic sequences on the premise of the global information of the previous round, while striving to decline the adverse effects of steganographic encoding on text quality. Moreover, for enhancing the semantic internal dependency of stegotexts, we utilize a constraint word sequences extraction scheme to obtain keywords to initialize the skeleton of targeted stegotexts, then expand the existing keywords with insertion operations. Experimental results demonstrate that PNG-Stega outperforms compared methods in terms of imperceptibility and anti-steganalysis ability. In particular, PNG-Stega provides high information hiding efficiency, even exceeding the autoregressive methods by around 2 times.

The current lexical substitution-based linguistic steganography primarily determines the substitutions by their linguistic suitability, overlooking the disparity in their capability against steganalysis. This oversight leads to the potential security risk by performing poor substitutions. To address this issue, this letter proposes aCausalPerceptionGuidedLinguisticSteganography(CPG-LS) via elaborate and secure lexical substitutions. CPG-LS constructs a causal perception network by making full use of a trained CNN discriminator to assess the security of each original word in the cover text and its substitutable candidates for controlling the embedding of secret message. Particularly, the causal score of each word is comprehensively measured from two perspectives by the causal perception network, one is word saliency, and the other is anti-steganalysis capability. Finally, the selection of original words is explicitly guided by their causal scores for securely embedding secret message. As the causal score of a word decreases, the perturbation caused by modifying this word becomes smaller, leading to stronger anti-steganalysis capability and lower semantic distortion in the obtained stegotexts. The experimental results demonstrate that CPG-LS achieves better text quality and anti-steganalysis capability than existing similar methods.

Abstract Deep learning approaches to interactive image segmentation are typically formulated as a binary labeling problem. A model trained to make predictions within a fixed set of labels (i.e., foreground and background labels) cannot be used to directly predict the binary masks of multiple objects of interest, which greatly limits its flexibility and adaptivity. The use of different classes of clicks as input is opted for and the first end‐to‐end learning model for multi‐object segmentation, based on a new designed neural network, is developed. The network consists of a visual feature extractor, a recurrent attention module and a dynamic segmentation head, extracts user click‐adapted appearance embedding features and spatial attention features, and then learns to transform this information into a segmentation of multiple objects. It is also proposed to train the network using a joint loss function, taking the embedding learning into account for segmentation. Comprehensive experiments are conducted on three benchmark datasets to demonstrate the effectiveness of the proposed method. It performs favorably against state‐of‐the‐art approaches on the multiple object segmentation task, for example, with 0.15 s per image, 0.06 s per object and mean IoU & F1 score of 84.90% on Pascal VOC 2012 validation set. It is further shown that the method can be used in numerous vision applications such as image recoloring and colorization.

Existing linguistic steganalysis methods share a similar conceptual learning paradigm: namely, learning specific features derived from a word and its surrounding words, then passing them to the classifier. However, these features are unable to deal with the impact of the association between words and the corpus, which gives rise to significant limitations. In this paper, we discover the missing link between the words and the corpus, and accordingly propose a framework of linguistic steganalysis named LS-BGAT. Specifically, we fine-tune a large-scale pre-training BERT as the local feature extraction model and employ Graph Attention Network (GAT) as the global feature extraction model. Through ensuring collaboration between the local BERT-based features and the global GAT-based features in the joint prediction layer, we combine rich semantic and syntactic information at the sentence level with underlying global information at the corpus level. Furthermore, we extend the binary classification of steganalysis to multi-category classification, thereby enhancing practicality. We empirically substantiate the effectiveness and universality of LS-BGAT on three tasks.

In recent years, linguistic generative steganography has been greatly developed. The previous works are mainly to optimize the perceptual-imperceptibility and statistical-imperceptibility of the generated steganographic text, and the latest developments show that they have been able to generate steganographic texts that look authentic enough. However, we noticed that these works generally cannot control the semantic expression of the generated steganographic text, and we believe this will bring potential security risks. We named this kind of security challenges as cognitive-imperceptibility. We think this is a new challenge that the generative steganography models must strive to overcome in the future. In this letter, we conduct some preliminary attempts to solve this challenge. Experimental results show that the proposed methods can further constrain the semantic expression of the generated steganographic text on the basis of ensuring certain perceptual-imperceptibility and statistical-imperceptibility, so as to enhance its cognitive-imperceptibility.

As a powerful learning tool, Extreme Learning Machine (ELM) shows its merits in classification, regression and clustering by offering both high prediction accuracy and high learning speed. Among numerous ELM varieties, multiple kernel ELM draws intensive attention from researchers because it can leverage information from multiple heterogeneous sources, which is a common scenario in big data era. Despite remarkable efforts for supervised multiple kernel ELM, few publications have addressed the unsupervised case, which is more critical yet challenging for tackling realistic problem. In this paper, we address this problem by proposing a two-stage unsupervised multiple kernel extreme learning machine, which is suitable for fast multiple-view clustering. This approach learns the cluster and kernel combination weights alternatively. At the first stage, it generates cluster label based on a given combined kernel. Then, at the second stage, the kernel combination weights are learned by distance label based extreme learning machine based on the label generated at the previous stage. Experimental results on both synthetic and real data sets demonstrate its outstanding performance in term of both accuracy and learning speed.

In the research area of text steganography, algorithms based on font format have advantages of great capacity, good imperceptibility and wide application range. However, little work on steganalysis for such algorithms has been reported in the literature. Based on the fact that the statistic features of font format will be changed after using font-format-based steganographic algorithms, we present a novel Support Vector Machine-based steganalysis algorithm to detect whether hidden information exists or not. This algorithm can not only effectively detect the existence of hidden information, but also estimate the hidden information length according to variations of font attribute value. As shown by experimental results, the detection accuracy of our algorithm reaches as high as 99.3% when the hidden information length is at least 16 bits.