Due to the rapid growth of machine learning tools and specifically deep networks in various computer vision and image processing areas, applications of Convolutional Neural Networks for watermarking have recently emerged. In this paper, we propose a deep end-to-end diffusion watermarking framework (ReDMark) which can learn a new watermarking algorithm in any desired transform space. The framework is composed of two Fully Convolutional Neural Networks with residual structure which handle embedding and extraction operations in real-time. The whole deep network is trained end-to-end to conduct a blind secure watermarking. The proposed framework simulates various attacks as a differentiable network layer to facilitate end-to-end training. The watermark data is diffused in a relatively wide area of the image to enhance security and robustness of the algorithm. Comparative results versus recent state-of-the-art researches highlight the superiority of the proposed framework in terms of imperceptibility, robustness and speed. The source codes of the proposed framework are publicly available at Github 1 .

ReDMark: Framework for Residual Diffusion Watermarking based on Deep Networks

This paper analyses several volatility models by examining their ability to forecast the Value-at-Risk (VaR) for two different time periods and two capitalization weighting schemes. Specifically, VaR is calculated for large and small capitalization stocks, based on Dow Jones (DJ) Euro Stoxx indices and is modeled for long and short trading positions by using non parametric, semi parametric and parametric methods. In order to choose one model among the various forecasting methods, a two-stage backtesting procedure is implemented. In the first stage the unconditional coverage test is used to examine the statistical accuracy of the models. In the second stage a loss function is applied to investigate whether the differences between the models, that calculated accurately the VaR, are statistically significant. Under this framework, the combination of a parametric model with the historical simulation produced robust results across the sample periods, market capitalization schemes, trading positions and confidence levels and therefore there is a risk measure that is reliable.

A Robust VaR Model under Different Time Periods and Weighting Schemes

A new multiplicative watermark detector in the contourlet domain using t Location-Scale distribution

Data security is a main concern in everyday data transmissions in the Internet. A possible solution to guarantee a secure and legitimate transaction is via hiding a piece of tractable information into the multimedia signal, i.e., watermarking. This brief proposes a new multiplicative image watermarking scheme in the contourlet domain by taking into account the local statistical properties and inter-scale dependencies of the contourlet coefficients of images. Although the contourlet coefficients are non-Gaussian within a sub-band, their local distribution fits the Gaussian distribution very well. In addition, it is known that there exist across-scale dependencies among these coefficients. In view of this, we propose the use of bivariate Gaussian (BVG) distribution to model the distribution of the contourlet coefficients. Motivated by the modeling results, an optimum blind watermark decoder is designed in the contourlet domain using the maximum likelihood method. By means of carrying out a number of experiments, the performance of the proposed decoder is investigated with regard to the bit error rate and compared to other decoders. It will be shown that the proposed decoder built upon the BVG model is superior to other decoders in terms of rate of error. It will also be shown that the proposed decoder provides higher robustness in comparison to other decoders in presence of attacks such as filtering, compression, cropping, scaling, and noise.

A Robust Image Watermarking Scheme Using Local Statistical Distribution in the Contourlet Domain

To enhance the invisibility and the robustness of the watermarking algorithm, a robust and imperceptible watermarking scheme is presented by combining Canny edge detection, contourlet transform with singular value decomposition. The host image is firstly decomposed by the contourlet transform. Then the low frequency sub-band obtained by contourlet transform is partitioned into 4×4 non-overlapping blocks and the singular value decomposition is carried out for the specific blocks selected by the Canny edge detection. Finally, the watermark is embedded into the coefficient of matrix U. The embedded watermark could be recovered blindly in the watermark extraction stage. And the robustness and the imperceptibility are efficiently guaranteed with an optimal threshold k selected by the least-square curve fitting. Experimental results demonstrate that the proposed watermarking scheme is superior in terms of imperceptibility and robustness against common attacks.

Robust and imperceptible watermarking scheme based on Canny edge detection and SVD in the contourlet domain

The RFM (Recency, Frequency and Monetary) model provides an effective analysis for decision makers in order to target their customers and develop appropriate marketing strategies according to their previous behaviors. Although the RFM model has been widely applied in various areas of marketing, its simplicity threatens its effectiveness since it does not consider the customers' relationship and changes in customers' behavior. In this paper, we propose an R+FM model which configures the segmentation according to the business changes and clusters customers using K-Means. We applied our model on Digikala company, the biggest E-Commerce in Middle East, and compared our model with the Digikala's previous RFM model which used Customer Quantile Method. Moreover, we built strategies for each segment and ran an SMS campaign according to those strategies. The results of the campaign showed that our Segmentation Model improved the number of purchase and average monetary of the baskets.

Customer Segmentation and Strategy Development Based on User Behavior Analysis, RFM Model and Data Mining Techniques: A Case Study

One application of digital watermarking is multimedia copyright protection. This paper address the blind watermark detection problem that copyright protection belongs to this field. Since watermark detection process can be modeled as a statistical decision rule, selecting appropriate statistical model plays an important role. This work presents a novel additive watermark detector for contourlet domain image watermarking. Contourlet coefficients of images have non-Gaussian property so, this property should be considered in modeling these coefficients. In this way, we use t location-scale distribution that have non-Gaussian property for modeling contourlet coefficients. Experimental results verify the high efficiency of the proposed detector, its higher performance in comparison with generalized Gaussian and Cauchy detector also its robustness against common image processing attacks such as image compression, median filtering and rotation.

Additive watermark detector in contourlet domain using the t location-scale distribution

Color texture image retrieval based on Copula multivariate modeling in the Shearlet domain

The use of renewable energy in power generation, sudden changes in load and fault in power transmission lines, causing a voltage drop in the system and challenge the reliability of the system. One way to compensate the changing nature of renewable energies in the short term without the need to disconnect loads or turn on other plants, use of renewable energies. The use of energy storage improved electrical stability, power quality and improve the peak power load. In this paper, using reinforcement learning to present optimal method for charge and discharge the consumer battery and Uncertainty of production could be due to the random nature of wind energy improved. Simulation results indicate not only the use of renewable energy and battery enhanced but also the cost of annual payments and peak consumption times reduced.

http://journals.iau.ir/article_519217_4696b42e1c93e9e4b76816c4ee209867.pdf

Sadegh Etemad

Papers

A new multiplicative watermark detector in the contourlet domain using t Location-Scale distribution

Customer Segmentation and Strategy Development Based on User Behavior Analysis, RFM Model and Data Mining Techniques: A Case Study

Additive watermark detector in contourlet domain using the t location-scale distribution

Color texture image retrieval based on Copula multivariate modeling in the Shearlet domain

Using reinforcement learning to make smart energy storage sources in microgrid