Search

Natural language generation (NLG) models combined with increasingly mature and powerful deep learning techniques have been widely used in recent years. Deployed NLG models in practical applications may be stolen or used illegally, and watermarking has become an important tool to protect the Intellectual Property (IP) of these deep models. Watermarking technique designs algorithms to embed watermark information and extracts watermark information for IP identification of NLG models can be seen as a symmetric signal processing problem. In terms of IP protection of NLG models, however, the existing watermarking approaches cannot provide reliable and timely model protection and prevent illegal users from utilizing the original performance of the stolen models. In addition, the quality of watermarked text sequences generated by some watermarking approaches is not high. In view of these, this paper proposes two embedding schemes to the hidden memory state of the RNN to protect the IP of NLG models for different tasks. Besides, we add a language model loss to the model decoder to improve the grammatical correctness of the output text sequences. During the experiments, it is proved that our approach does not compromise the performance of the original NLG models on the corresponding datasets and outputs high-quality text sequences, while forged secret keys will generate unusable NLG models, thus defeating the purpose of model infringement. Besides, we also conduct sufficient experiments to prove that the proposed model has strong robustness under different attacks.

Deep neural networks have achieved remarkable success in various fields of artificial intelligence. However, these models, which contain a large number of parameters, are widely distributed and disseminated by researchers, engineers, and even unauthorized users. Except for intelligent tasks, typically overparameterized deep neural networks have become new digital covers for data hiding, which may pose significant security challenges to AI systems. To address this issue, this paper proposes a symmetric steganalysis scheme specifically designed for neural networks trained for image classification tasks. The proposed method focuses on detecting the presence of additional data without access to the internal structure or parameters of the host network. It employs a well-designed method based on histogram distribution to find the optimal decision threshold, with a symmetric determining rule where the original networks and stego networks undergo two highly symmetrical flows to generate the classification labels; the method has been shown to be practical and effective. SVM and ensemble classifiers were chosen as the binary classifier for their applicability to feature vectors output from neural networks based on different datasets and network structures. This scheme is the first of its kind, focusing on steganalysis for neural networks based on the distribution of network output, compared to conventional digital media such as images, audio, and video. Overall, the proposed scheme offers a promising approach to enhancing the security of deep neural networks against data hiding attacks.

This paper proposes a repeatable data hiding framework for digital images, in which the distortion caused by data hiding is invariable no matter how many times the embedding operation is executed. As a result, the usability of images can be always guaranteed. To achieve repeatable data hiding, we deduce the theoretical modification probabilities of the elements in cover image (an image used for data hiding). Then we design an embedding framework to make the practical modification probabilities of cover elements equal to the deduced ones. When additional data is embedded into cover image with the deduced probabilities, the modification-trace of current embedding is replaced by the latter embedding. That means multiple embeddings do not produce new modification-trace, so that the distortion caused by data hiding is invariable. Therefore, the repeatability of data hiding can be guaranteed. Finally, we describe some applications of repeatable data hiding, e.g., logistics management, authentication database management, and steganography, to show the practicability of our framework.

Recently, neural networks have become a promising architecture for some intelligent tasks. In addition to conventional tasks such as classification, neural networks can be used for data hiding. This paper proposes a data hiding scheme to transmit different data to multiple receivers via the same neural network simultaneously. Additional data are embedded into the neural network during the process of training instead of modifying the network parameters after training. As a result, the reduction in detection accuracy of the neural network is trivial with additional data embedded. On the receiver side, additional data can be extracted using specific data decoding networks. Instead of training, the parameters of data decoding networks are generated using the embedding keys possessed by the receivers. Therefore, it is unnecessary to store and transmit the decoding networks secretly, and the practicability is satisfactory. A receiver can extract corresponding additional data with the correct embedding key. For other parts of additional data, the receiver cannot affirm their existence and therefore cannot extract them. Experimental results verify the effectiveness of the proposed data hiding scheme, including embedding capacity and robustness.

Batch steganography is a technique that hides information into multiple covers. To achieve a better performance on the security of data hiding, we propose a novel strategy of batch steganography using a generative network. In this method, the approaches of cover selection, payload allocation, and distortion evaluation are considered in the round. We define a quality metric to evaluate the distortion between the cover image and the stego. When training the generation function, we define an objective function containing two parts: the entropy loss and the steganalytic loss. While the entropy loss is used to represent the gap between the payload inside stego images and the entire embedding capacity, the steganalytic loss is used to assess the data embedding impact using the proposed quality metric. With back-propagation, we minimize the objective function to obtain an optimal solution. Accordingly, different payloads can be allocated to different images, and the ± 1 modification probability for pixels in each cover can be calculated. Finally, we embed information into the selected images by STC. Experimental results show that the proposed method achieves a better undetectability against modern steganalytic tools.

In this article, we focus on a new concept called multichannel steganography, in which a sender is able to transmit different secret data to multiple receivers via the same cover image. First, we propose a general framework to achieve multichannel steganography based on modern steganographic paradigm. Then, we design two schemes to, respectively, achieve separable and sequential multichannel steganography with the proposed framework. In the separable scheme, a receiver can extract secret data for him/her using the corresponding data hiding key. For other parts of secret data, the receiver cannot affirm the existence and, therefore, cannot extract them. In the sequential scheme, the multiple receivers are dependent. The receiver with the highest priority can extract data first. After that, the receiver with the second priority can extract data when authorized by the highest priority receiver, and so on. In addition, the two schemes achieve multichannel steganography without decreasing the undetectability of steganography.

Content similarity is a representative property of natural images, for example, similar regions, which is utilized by modern steganalysis. Existing JPEG steganographic methods mainly focus on the complexity of content but ignore content similarity. This article investigates content similarity to improve the undetectability of JPEG steganography. Specifically, the content similarity of DCT blocks and the 64 parallel channels is used to design the distortion function. Given a JPEG image, initial embedding costs are assigned for quantized DCT coefficients using an appropriate algorithm among the existing distortion functions. Then, the similarities of blocks and channels are used to update the initial embedding costs, respectively. After combination, the final distortion function can be obtained. Using syndrome trellis coding (STC), which achieves minimal embedding distortion with respect to a given distortion function, secret data are embedded into the cover image with a final distortion function. Experimental results show that our scheme achieves better undetectability than current state-of-the-art JPEG steganographic methods.

This paper proposes a new steganography method for hiding data into dynamic GIF (Graphics Interchange Format) images. When using the STC framework, we propose a new algorithm of cost assignment according to the characteristics of dynamic GIF images, including the image palette and the correlation of interframes. We also propose a payload allocation algorithm for different frames. First, we reorder the palette of GIF images to reduce the modifications on pixel values when modifying the index values. As the different modifications on index values would result in different impacts on pixel values, we assign the elements with less impact on pixel values with small embedding costs. Meanwhile, small embedding costs are also assigned for the elements in the regions that the interframe changes are large enough. Finally, we calculate an appropriate payload for each frame using the embedding probability obtained from the proposed distortion function. Experimental results show that the proposed method has a better security performance than state-of-the-art works.

In order to seek the coordination of energy supply system in economy, security, environmental protection and energy conservation, relying on the "Internet plus", a distributed cooling, heating and power system integrated with power generation unit (PGU), ground source heat pump and energy storage system was constructed. The models about off-design condition of each device were illustrated. On this basis, four operation strategies were put forward: following electric load, following thermal load, following hybrid load and following max efficiency of PGU. Then operation cost, carbon dioxide emission and primary energy were employed as objectives to analyze the operating time, output and performances of the proposed system under off-design condition. The results show that the proposed system has better economy, lower emissions and lower primary energy consumption than the conventional separate system. Under FTL strategy, the operation cost and carbon dioxide emission are the lowest, which are reduced by 32.7% and 45.3%, respectively. Because of the continuous operating, the reduction of primary energy is the most, which is 86.7%. The results can provide some theoretical references for the operation of distributed system.

With the development of smartphones and mobile wireless network, social discovery applications are becoming more and more popular now. These applications help users to chat with nearby people and m...

This paper proposes a secure steganographic method in social networks based on the behavioral correlation between the sender and his friends. A sender transmits secret data to one of his friends in...

This letter focus on a practical scenario of cover selection for steganography, in which a part of the available images of the sender have been processed to improve visual quality, e.g., contrast enhancement, image denoising. In this case, not only the embedding distortion (caused by steganography), but also the processing distortion (caused by processing) should be considered when selecting cover image. We propose a cover selection method to combine the two kinds of distortion together to measure the suitability for steganography of the available images. To calculate the processing distortion, a classifier is trained to distinguish the processed images with the original ones. The classifier is then used to measure the possibility of the existence of processing. High possibility means high processing distortion. In addition, the current distortion minimization framework designed for steganography is employed to calculate the embedding distortion. Finally, both kinds of distortion are combined to form the total distortion, and the image with the minimal total distortion is selected as cover. Using the selected image for embedding, high undetectability can be achieved.

This paper proposes a novel system to protect the fingerprint database based on compressed binary fingerprint images. In this system, the user can store private data with high capacity. First, the grayscale fingerprint image is transformed into a binary bitstream. Then, the binary bitstream is compressed using run-length encoding and Huffman encoding technique to create a sparse space to accommodate private data. Finally, the new image constructed by the obtained binary bitstream is encrypted with an encryption key. For fingerprint matching, it is only need to decrypt and decompress the encrypted image in the database to obtain the binary fingerprint image. When the matching is passed, the private data can be extracted to a user with data-embedding key. If the matching is failed, the private data cannot extracted for the purpose of privacy security. Even if a leakage of the encryption key occurs, this system can still protect the privacy data of the user due to the existence of the data-embedding key. Meanwhile, the encoding and decoding phases are real time, which guarantee the practicability of the proposed system.

Posting images on social network platforms is happening everywhere and every single second. Thus, the communication channels offered by various social networks have a great potential for covert communication. However, images transmitted through such channels will usually be JPEG compressed, which fails most of the existing steganographic schemes. In this paper, we propose a novel image steganography framework that is robust for such channels. In particular, we first obtain the channel compressed version (i.e., the channel output) of the original image. Secret data is embedded into the channel compressed original image by using any of the existing JPEG steganographic schemes, which produces the stego-image after the channel transmission. To generate the corresponding image before the channel transmission (termed the intermediate image), we propose a coefficient adjustment scheme to slightly modify the original image based on the stego-image. The adjustment is done such that the channel compressed version of the intermediate image is exactly the same as the stego-image. Therefore, after the channel transmission, secret data can be extracted from the stego-image with 100% accuracy. Various experiments are conducted to show the effectiveness of the proposed framework for image steganography robust to JPEG compression.

This article describes how to overcome the shortage of steganalysis for small capacity-based embedding. A slight modification method is proposed to break steganography. For a given image, traditional steganalysis methods are first used to achieve a preliminary result. For the “clear” image judged by steganalysis, it is still suspicious because of the incompleteness of steganalysis for small capacity. Thus, slight modifications are made to break the possibility of covert communication. The modifications are made on the locations with minimal distortion to guarantee high quality of the modified image. To this end, a proposed distortion minimization based algorithm using slight modification. Experimental results show that the error rate of secret data extraction is around 50% after implementation, which indicates that the covert communication of steganography is destroyed completely.

This article describes how the existing distortion functions for JPEG steganography allot same cost for ±1 embedding changes. Because of the correlation of natural image, however, changes with different polarities make different influences on image. Therefore, the embedding costs for ±1 embedding changes should not be equivalent. This article proposes a general method to distinguish the embedding costs for different polarities of embedding changes for JPEG images with the help of reference images constructed by block artifact compensation. The original JPEG image is decompressed into spatial domain firstly, and then the block artifact is compensated by smoothing filtering implemented on border pixels of each 8 × 8 block. After that, the compensated image which is more similar to the original uncompressed image is recompressed into DCT domain and adopted as side information to guide the adjusting of the given distortion function. Experiment results show that after the proposed method is employed, the security performance of current popular JPEG steganographic methods is observably increased.

Existing cover selection methods for steganography cannot resist pooled steganalysis. This paper proposes a secure cover selection method which is able to resist pooled steganalysis and single object steganalysis meanwhile. To resist pooled steganalysis, the maximum mean discrepancy (MMD) distance between the stego set and a clear arbitrary image set is kept not larger than a normal threshold during cover selection, where the threshold is the MMD distance between two clear arbitrary image sets. Under this constraint, a searching strategy is designed to select the minimal steganographic distortion images within the affordable computational complexity to resist single object steganalysis. With the selected covers, the security of steganography is guaranteed against both single object steganalysis and pooled steganalysis. The experimental results demonstrated the effectiveness of the proposed method.

Existing distortion functions in steganography which achieved high undetectability are designed for unprocessed natural image. Nowadays, a large number of images are filtered before transmitting for the sake of beautification. In this situation, existing distortion functions should be improved to fit the properties of these beautified images. This paper proposes a distortion function optimization method for steganography on beautified images. Given an unprocessed image, a popular image beautification software is employed to produce two similar beautified images. One of them is used for embedding and the other one is employed as reference. Guided by the reference, existing distortion functions are improved by distinguishing the embedding costs for ±1 embedding. After embedding, the stego image is closer to the reference, which results in a higher undetectability against steganalysis. Experimental results also proved the increasing of undetectability when examined by modern steganalytic tools.

At present, the loading speed of model scene is slow in the visual operation and maintenance process of substation GIS equipment, which leads to poor visual operation and maintenance effect. In order to solve this problem, a remote visual operation and maintenance method of substation GIS equipment based on digital modeling is proposed. According to IEC61850 standard, the digital modeling of substation AC charging pile, DC charger and other main equipment is completed. The multi tree hybrid structure is generated by using the model data organization and dynamic operation and maintenance method, and the ground object model of substation is constructed to improve the loading speed of the model. The remote visual operation and maintenance of substation GIS equipment is realized by using the map model mapping technology. The experimental results show that the method can effectively improve the loading speed of the model scene, the average loading frame rate is 42 frame/s, and has good visual operation and maintenance effect and strong practical application.

International audience; With the fast development of cloud computing, exchanging JPEG compressed images in a secure way has significantly increased. Data hiding in encrypted images (DHEI) is an effective way to embed additional data directly into the encrypted domain. In recent state-of-the-art methods, almost all DHEI processes focused on uncompressed images. Recently, some schemes using data hiding (DH) in JPEG crypto-compressed images have been designed, but most of them are not fully JPEG format compliant. In this paper, we propose a hierarchical high capacity data hiding (HHCDH) approach for JPEG cryptocompressed images. After encrypting every non-null coefficients, they are processed from low to high frequencies. Sign bits that are specific to them are then substituted by bits of a secret message. During the decoding phase, correlations between neighboring blocks are exploited to hierarchically recover the original sign bit values. According to our experiments, we achieve to obtain a high payload value, while preserving a very good quality of the reconstructed JPEG image. Index Terms-Signal processing in the encrypted domain, data hiding, crypto-compression, JPEG compression.

The deep learning based image steganalysis is becoming a serious threat to modification-based image steganography in recent years. Generation-based steganography directly produces stego images with secret data and can resist the advanced steganalysis algorithms. This paper proposes a novel generation-based steganography method by disguising the stego images into the kinds of images processed by normal operations (e.g., histogram equalization and sharpening). Firstly, an image processing model is trained using DCGAN and WGAN-GP, which is used to generate the images processed by normal operations. Then, the noise mapped by secret data is inputted into the trained model, and the obtained stego image is indistinguishable from the processed image. In this way, the steganographic process can be covered by the process of image processing, leaving little embedding trace in the process of steganography. As a result, the security of steganography is guaranteed. Experimental results show that the proposed scheme has better security performance than the existing steganographic methods when checked by state-of-the-art steganalytic tools, and the superiority and applicability of the proposed work are shown.

Contrast enhancement is widely used to improve the visual quality of images. This paper proposes a distortion function for steganography in enhanced images. The pixel prediction error and the cost of the pixel value itself are joint to fit the unique properties of enhanced image. Given a cover image, each pixel is predicted by its neighbors. The prediction error, which reflects the texture complexity, are combined with the cost of the pixel value itself to form the final distortion function. With the proposed distortion function, secret data can be embedded into the enhanced cover image with minimal image distortion using syndrome trellis coding (STC). As a result, less detectable artifacts left in the stego images, so that high undetectability is achieved.

In this paper, we propose a privacy protection scheme using image dual-inpainting and data hiding. In the proposed scheme, the privacy contents in the original image are concealed, which are reversible that the privacy content can be perfectly recovered. We use an interactive approach to select the areas to be protected, that is, the protection data. To address the disadvantage that single image inpainting is susceptible to forensic localization, we propose a dual-inpainting algorithm to implement the object removal task. The protection data is embedded into the image with object removed using a popular data hiding method. We further use the pattern noise forensic detection and the objective metrics to assess the proposed method. The results on different scenarios show that the proposed scheme can achieve better visual quality and antiforensic capability than the state-of-the-art works.

In this paper, we propose a new data-hiding framework: multisource data hiding, in which multiple senders (multiple sources) are able to transmit different secret data to a receiver via the same cover image symmetrically. We propose two multisource data-hiding schemes, i.e., separable and anonymous, according to different applications. In the separable scheme, the receiver can extract the secret data transmitted by all senders using the symmetrical data-hiding key. A sender is unable to know the content of the secret data that is not transmitted by them (non-source sender). In the anonymous scheme, it is unnecessary to extract all secret data on the receiver side. The content extracted by the receiver is a co-determined result of the secret data transmitted by all senders. Details of the secret data are unknown to the receiver and the non-source senders. In addition, the two proposed schemes achieve multisource data hiding without decreasing the undetectability of data hiding.

This paper proposes an idea of data computing in the covert domain (DCCD). We show that with information hiding some data computing tasks can be executed beneath the covers like images, audios, random data, etc. In the proposed framework, a sender hides his source data into two covers and uploads them onto a server. The server executes computation within the stego and returns the covert computing result to a receiver. With the covert result, the receiver can extract the computing result of the source data. During the process, it is imperceptible for the server and the adversaries to obtain the source data as they are hidden in the cover. The transmission can be done over public channels. Meanwhile, since the computation is realized in the covert domain, the cloud cannot obtain the knowledge of the computing result. Therefore, the proposed idea is useful for secure computing.

This letter proposes a batch steganographic method, which combines cover-selection and payload-allocation by steganographic distortion optimization. We first proved that with the value of payload increasing, the first-order derivative of steganographic distortion of a single cover is monotonically increasing. Then, we deduced that the first-order derivative of steganographic distortion of covers that selected from a given set should be equal if the total steganographic distortion of the corresponding selected covers is minimal. Finally, an algorithm was designed to combine cover-selection and payload-allocation, so that the optimal stego-covers can be obtained. Experiment results show that the undetectability is obviously improved when using the proposed steganographic method.

This paper proposes a framework to improve the existing distortion functions designed for JPEG steganography, which results in a better capability of countering steganalysis. Different from the existing steganography approach that minimizes image distortion, we minimize the feature distortion caused by data embedding. Given a JPEG image, we construct a reference image close to the image before JPEG compression. Guided by both the reference image and the feature distortion minimization, the state-of-the-art distortion functions designed for syndrome trellis coding embedding are improved by distinguishing the embedding costs for +1 versus -1 embedding. This paper has three contributions. First, the proposed framework outperforms the traditional, since we use the constructed reference image and the public steganalytic knowledge for data embedding. Second, the proposed framework is universal for improving distortion functions that were designed for JPEG steganography. Finally, experimental results also prove that the proposed approach has a better undetectability when examined by modern steganalytic tools.

Steganography aims to embed additional data into digital media secretly and minimize the distortion caused by data embedding. Up to now, the key point of minimal distortion embedding framework is to design proper distortion function. In this paper, a distortion function for JPEG steganography based on residual blocks is proposed. To obtain less statistical detectability, Residual Block Values (termed as RBVs for short) and quantization steps are both involved in the proposed distortion function. RBVs are exploited to determine the embedding risk caused by modifications in corresponding DCT block while quantization steps are exploited to determine the embedding risk of the selection channel in a block. By implementing the syndrome trellis coding (STC) to embed secret data, the modifications are constrained in hard-to-detect religions. Experiments show that the proposed method performs better than current state-of-the-art methods of JPEG steganography.

This paper proposes a novel distortion function for JPEG steganography, which depends on the magnitude of discrete cosine transformation (DCT) coefficients, the first- and second-order residuals, a...

This paper proposes an improved steganalytic method when cover selection is used in steganography. We observed that the covers selected by existing cover selection methods normally have different characteristics from normal ones, and propose a steganalytic method to capture such differences. As a result, the detection accuracy of steganalysis is increased. In our method, we consider a number of images collected from one or more target (suspected but not known) users, and use an unsupervised learning algorithm such as $k$ -means to adapt the performance of a pre-trained classifier towards the cover selection operation of the target user(s). The adaptation is done via pseudo-labels from the suspected images themselves, thus allowing the re-trained classifier more aligned with the cover selection operation of the target user(s). We give experimental results to show that our method can indeed help increase the detection accuracy, especially when the percentage of stego images is between 0.3 and 0.7.

Cooperating with secondary protection system, DC circuit breakers are used to deal with direct current (DC) faults in low voltage DC distribution system. To realize ultra-fast operating of DC breakers, optimization based on simulation of coil-coil type electromagnetic repulsion actuator (EMRA) is discussed in this paper. Considering exiting simulation based on finite element method (FEM) is quite time-consuming, an equivalent circuit method (ECM) for the Coil-Coil type EMRA is established. In the equivalent circuit equation, lumped equivalent inductance which contributes to the induced electromotive force (EMF) is used to reflect the coupling effect of distribute electromagnetic field on electric circuit, the gradient of equivalent inductance which contributes to the motional EMF reflecting the coupling effect of mechanical motion on electric circuit and the equivalent resistance that varies with temperature reflecting the coupling effect of temperature field on electric circuit. The lumped equivalent inductance and its gradient are extracted by solving static magnetic field based on FEM. By solving the state equations, the transient electrical and operating characteristic of EMRA are obtained. In the end, the results of the model based on ECM are compared with FEM, and as expected, they are in good consistence.

The working principle of the dual active bridge DC/DC converter is analyzed, and the equivalent circuit models of DC/DC converter are studied when pole-to-pole fault and pole-to-ground fault occur in the high-voltage side. When the system has pole-to-pole fault on the high-voltage side of the DC/DC converter, the transient characteristics of electric quantities are studied. When the system has pole-to-ground fault on the high-voltage side of the DC/DC converter, the transient characteristics of the electrical quantities are studied in the case of converter blocks immediately and does not block, respectively. The simulation model is built using PSCAD/EMTDC to verify the theoretical analysis results.

This paper proposes a new feature extraction method for steganalysis. In the improved method, the feature set is equipped with two parts. The first part is obtained by employing an existing feature extraction method on a suspicious image. With a content-correlative operation, differences of the image features are enlarge between the cover and the stego. The second part is obtained by employing the same feature extraction method on the obtained image. The two parts are combined together to form the new feature set. With the new feature set, the modifications made by steganography can be recognized more efficiently. Experimental results show that the detection accuracy of existing feature extraction methods is increased using the proposed method.

This paper considers a more practical situation for pooled steganalysis that only a single actor is observed, so that the steganalyst needed to analyze the actor independently without comparing with other actors. We propose a pooled steganalysis method for this situation. For a given actor that has emitted a number of images, feature sets are extracted from each image, respectively, and then feed to a binary classifier popularly used in single object steganalysis. Combining all the results output by the classifier, a final decision is made ensemble to label the given actor as “guilty” or “innocent” with the minimal detection error. Experimental results show that the proposed method is effective for single actor pooled steganalysis.

With the rapid development of deep learning, a lot of CNN-based steganalyzers have emerged. This kind of steganalyzer uses statistical learning to investigate the properties caused by steganography, which is the most efficient approaches for breaking information hiding. However, we find a vulnerability of CNN-based steganalyzer that it can be defeated by dual operations. In this paper, we propose an easy yet effective algorithm to perturb the stego images against neural network, which can evade CNN-based steganalyzer with high probabilities. We elaborated on the theoretical basis of the method we proposed and proved the feasibility of this method through experiments.

International audience; Nowadays, emoji image is widely used in social networks. To achieve covert communication in emoji images, this paper proposes a distortion function for emoji images steganography. The profile of image content, the intra-and inter-frame correlation are taken into account in the proposed distortion function to fit the unique properties of emoji image. The three parts are combined together to measure the risks of detection due to the modification on the cover data. With the popular syndrome trellis coding (STC), the distortion of stego emoji image is minimized using the proposed distortion function. As a result, less detectable artifacts could be found in the stego images. Experimental results show that the proposed distortion function performs much higher undetectability than current state-of-the-art distortion function HILL which is designed for natural image.