One of the open problems in neural network research is how to automatically determine architectures for given applications. In this brief, we propose a simple and efficient approach number hidden nodes generalized single-hidden-layer feedforward networks (SLFNs) which need not be alike. This referred as error minimized extreme learning machine (EM-ELM) can add random SLFNs one by or group (with varying size). During growth networks, output weights are updated incrementally. The convergence...

This paper proposes two novel self-embedding watermarking schemes based upon a reference sharing mechanism, in which the watermark to be embedded is derived from original principal content different regions and shared by these for restoration. After identifying tampered blocks, both data reserved area are used recover area. By using first scheme, five most significant bit layers of cover image can recovered watermarked also retrieved when replacement not too extensive. In second host...

This paper proposes a novel watermarking scheme with flexible self-recovery quality. The embedded watermark data for content recovery are calculated from the original discrete cosine transform (DCT) coefficients of host image and do not contain any additional redundancy. When part watermarked is tampered, in area without modification still can be extracted. If amount extracted large, we reconstruct tampered according to constraints given by data. Otherwise, may employ compressive sensing...

This paper proposes a method of reversible data hiding in encrypted images (RDH-EI) based on progressive recovery. Three parties are involved the framework, including content owner, data-hider, and recipient. The owner encrypts original image using stream cipher algorithm uploads ciphertext to server. data-hider server divides into three channels and, respectively, embeds different amount additional bits each one generate marked image. On recipient side, message can be extracted from image,...

With the rise of cryptocurrencies, blockchain, which is underlying technology them, has gained more attention and been used in Internet Things (IoT) other fields. However, there are bottlenecks that hinder its application, such as storage capacity. Due to large number IoT devices always act data generators many systems, transactions will be generated at a high rate. The problem serious IoT. In this article, expand capacity for each peer, we select old blocks created previously less likely...

In this paper, a novel perceptual image hashing scheme based on convolutional neural network (CNN) with multiple constraints is proposed, in which our deep learns the process of features extraction automatically according to training target and then generates final hash sequence. The combination pooling layers reduce size input while deepening channels. Then, we construct two pairs integrate them into an overall constraint function through strategy weight allocation. order guarantee...

Multi-view clustering aims at simultaneously obtaining a consensus underlying subspace across multiple views and conducting on the learned subspace, which has gained variety of interest in image processing. In this paper, we propose Semi-supervised Structured Subspace Learning algorithm for data points from Multiple sources (SSSL-M). We explicitly extend traditional multi-view with semi-supervised manner then build an anti-block-diagonal indicator matrix small amount supervisory information...

Multiview clustering has received great attention and numerous subspace algorithms for multiview data have been presented. However, most of these do not effectively handle high-dimensional fail to exploit consistency the number connected components in similarity matrices different views. In this article, we propose a novel consistency-induced (CiMSC) tackle issues, which is mainly composed structural (SC) sample assignment (SAC). To be specific, SC aims learn matrix each single view wherein...

Watermarking by Zernike moments has been proven to be effective in providing high rotational resistance. However, due the computational complexity, conventional video watermarking methods using are developed for videos with low resolution. Moreover, according properties of moments, only matrices equal height and width can calculated since inscribed circle original image matrix is selected as area processed, but most available on Internet do not meet such requirement. To solve above problem,...

Multi-view clustering has gained great progress recently, which employs the representations from different views for improving final performance. In this paper, we focus on problem of multi-view based Markov chain by considering low-rank constraints. Since most existing methods fail to simultaneously characterize relations among entries in a tensor global perspective and describe local structures similarity matrices tensor, propose novel Flexible Tensor Learning Clustering with (FTLMCM)...

This paper proposes a novel scheme of compressing encrypted images with auxiliary information. The content owner encrypts the original uncompressed and also generates some information, which will be used for data compression image reconstruction. Then, channel provider who cannot access may compress by quantization method optimal parameters that are derived from part information ratio-distortion criteria, transmit compressed data, include an sub-image, quantized another At receiver side,...

Digital watermark embeds information bits into digital cover such as images and videos to prove the creator's ownership of his work. In this paper, we propose a robust image algorithm based on generative adversarial network. This model includes two modules, generator adversary. Generator is mainly used generate embedded with watermark, decode damaged by noise obtain watermark. Adversary discriminate whether damage noise. Based Hidden (hiding data deep networks), add high-pass filter in front...

Steganalysis is a technique for detecting the existence of secret information hidden in digital media. In this paper, we propose novel scheme JPEG steganalysis. scheme, first design diverse base filters which are able to obtain image residuals from various directions. Then, cascade filter generation strategy construct set high order filters. We further select with maximum diversity. The selected convolved decompressed capture subtle embedding traces. residuals, termed as diversity residual,...

Recent linguistic steganalysis methods model texts as sequences and use deep learning models to extract discriminative features for detecting the presence of secret information in texts. However, natural language has a complex syntactic structure have limited representation ability text modeling. Moreover, previous tend from local continuous word sequences, which cannot effectively global characteristics. In this paper, we present method with graph neural network. proposed method, are...

Linguistic steganography (LS) aims to embed secret information into a highly encoded text for covert communication. It can be roughly divided two main categories, i.e., modification based LS (MLS) and generation (GLS). MLS embeds data by slightly modifying given without impairing the meaning of text, whereas GLS uses well trained language model directly generate carrying data. A common disadvantage methods is that embedding payload very small, whose return preserving semantic quality text....

This work proposes a novel scheme of compressing and decompressing encrypted image based on compressive sensing. An original is as set coefficients by secret orthogonal transform. Since the has sparse representation in conventional transform domain can be recovered from small quantity measurements, data are compressed into series measurement data. Using signal recovery method sensing, receiver reconstruct principal content image. way, quality reconstructed dependent compression rate...

Extreme learning machine (ELM), proposed by Huang et al., was developed for generalized single hidden layer feedforward networks with a wide variety of nodes. ELMs have been proved very fast and effective especially solving function approximation problems predetermined network structure. However, it may contain insignificant In this paper, we propose dynamic adjustment ELM (DA-ELM) that can further tune the input parameters nodes in order to reduce residual error. It is paper energy error be...