Landslide-Dammed Mapping and Logistic Regression Modeling Using GIS and R Statistical Software in the Northeast Afghanistan

A complex terrain and topography resulted in an enormous landslide-dammed area northeast of Afghanistan. Moreover, debris, rock avalanches, and landslides occurrences are the primary source of lakes created within the area. Recently, instances have increased because of the high displacement and mass movement by glacial and seismic activities. In this study, using GIS and R statistical software, we performed a logistic regression modeling in order to map and predict the probability of landslides-dammed occurrences. Totally, 361 lakes were mapped using Google Earth historical imagery. This total was divided into 253 （70%） lakes for modeling and 801 （30%） lakes for the model validation. They were randomly selected by creating a fishnet for the study area using Arc toolbox in GIS. Four independent variables that are mostly contributed to landslide-dammed occurrences consisting of slope angles, relief classes, distances to major water sources and earthquake epicenters, were extracted from DEM （digital elevation model） data using 85-meter resolution. The result is a grid map that classified the area into Low （16,834.98 km2）, Medium （2,217.302 kin：） and High （2,013.55 km2） vulnerability to landslide-dammed occurrences. Overall, the model result has been validated by using a ROC （receiver operator characteristic） curve available in SPSS software. The model validation showed a 95.1 percent prediction accuracy that is considered satisfactory.

An Efficient Information Hiding Scheme Based on Closest Paired Tree Structure Vector Quantization

Information hiding schemes based on vector quantization （VQ） usually require lengthy VQ encoding and decoding processes. In this paper, we propose an efficient information hiding method based on closest paired tree structure vector quantization （CPTSVQ）. The simulation result shows that the execution time of the proposed scheme is much shorter than that attained by previous approaches.

Study of Difference-Expansion Embedding Schemes

Difference expansion（DE） is one of the famous schemes in the field of reversible data hiding.With the high efficiency and simplicity,DE also has received more attention over the years.DE has a good information capacity,but due to its major location map,the pure payload is rather low.Therefore many scholars did relevant improvements which let n pixels as a unit instead of the original two pixels as a unit and can adaptively adjust the number of embedding secret information according to the smoothness degree of the block,which achieves the result of improving the information payload or the image quality.In this paper,the study of DE-based reversible data hiding schemes is comprehensively discussed.The performance of DEbased schemes is evaluated and compared in terms of embedding capacity and stego-image quality.

Unsupervised Electric Motor Fault Detection by Using Deep Autoencoders

Fault diagnosis of electric motors is a fundamental task for production line testing, and it is usually performed by experienced human operators. In the recent years, several methods have been proposed in the literature for detecting faults automatically. Deep neural networks have been successfully employed for this task, but, up to the authors' knowledge, they have never been used in an unsupervised scenario. This paper proposes an unsupervised method for diagnosing faults of electric motors by using a novelty detection approach based on deep autoencoders. In the proposed method, vibration signals are acquired by using accelerometers and processed to extract LogMel coefficients as features. Autoencoders are trained by using normal data only, i.e., data that do not contain faults. Three different autoencoders architectures have been evaluated: the multilayer perceptron(MLP) autoencoder, the convolutional neural network autoencoder, and the recurrent autoencoder composed of long short-term memory(LSTM) units. The experiments have been conducted by using a dataset created by the authors, and the proposed approaches have been compared to the one-class support vector machine(OC-SVM) algorithm. The performance has been evaluated in terms area under curve(AUC) of the receiver operating characteristic curve, and the results showed that all the autoencoder-based approaches outperform the OCSVM algorithm. Moreover, the MLP autoencoder is the most performing architecture, achieving an AUC equal to 99.11 %.

A Road Extraction Method for Remote Sensing Image Based on Encoder-Decoder Network

According to the characteristics of the road features,an Encoder-Decoder deep semantic segmentation network is designed for the road extraction of remote sensing images.Firstly,as the features of the road target are rich in local details and simple in semantic features,an Encoder-Decoder network with shallow layers and high resolution is designed to improve the ability to represent detail information.Secondly,as the road area is a small proportion in remote sensing images,the cross-entropy loss function is improved,which solves the imbalance between positive and negative samples in the training process.Experiments on large road extraction datasets show that the proposed method gets the recall rate 83.9%,precision 82.5%and F1-score 82.9%,which can extract the road targets in remote sensing images completely and accurately.The Encoder-Decoder network designed in this paper performs well in the road extraction task and needs less artificial participation,so it has a good application prospect.

Distortion-Free Data Embedding Scheme for High Dynamic Range Images

Distortion-free data embedding is a technique which can assure that not only the secret data is correctly extracted but also the cover media is recovered without any distortion after secret data is extracted completely. Because of these advantages, this technique attracts the attention of many researchers. In this paper, a new distortion-free data embedding scheme for high dynamic range （HDR） images is proposed. By depending on Cartesian product, this scheme can obtain higher embedding capacity while maintaining the exactly identical cover image and stego image when using the tone mapping algorithms. In experimental results, the proposed scheme is superior to Yu et aL＇s scheme in regard to the embedding rate——an average embedding rate of 0.1355 bpp compared with Yn et aL＇s scheme （0.1270 bpp）.

A Visual Secret Sharing Scheme for Progressively Restoring Secrets

Visual secret sharing has received more and more attention over the past years due to the fact that neither complex computation nor cryptographic knowledge is required to decrypt the secret image directly according to the characteristics of the human vision system. Considering the issue of sharing the secret image at multiple image resolutions with the meaningful shadows, in this paper, we present a friendly progressive visual secret sharing scheme without expanding the image size in the shadows and the reconstructed secret image based on applying a 2x2-sized block-wise operation to generate the shadows block by block. The proposed method can achieve these benefits： 1） the generated shadows are meaningful, and their sizes are not expanded, and 2） the secret image can be recovered at different resolutions by stacking different quantities of shadows together. The experimental results also show that the proposed method is superior to other compared schemes.

A Novel(n,n)Secret Image Sharing Scheme Based on Sudoku

Recently, Chang et al. proposed a Sudoku-based secret image sharing scheme. They utilized the Sudoku grid to generate meaningful shadow images, and their scheme satisfied all essential requirements. Based on Chang et al.＇s scheme, we propose a novel （n, n） secret image sharing scheme based on Sudoku. In the proposed scheme, a secret image can be shared among n different cover images by generating n shadow images, and the secret image can be reconstructed without distortion using only these n shadow images. Also, the proposed scheme can solve the overflow and underflow problems. The experimental results show that the visual quality of the shadow images is satisfactory. Furthermore, the proposed scheme allows for a large embedding capacity.

A Distributed Reactive Power Sharing Approach in Microgrids with Improved Droop Control

Conventional droop control causes frequency and voltage deviations(from rated value)in a inverter-intensive microgrid(MG),and the reactive power sharing cannot be obtained when the communication structure of the MG or load suddenly changes.Compared with a centralized control and droop control scheme,a distributed hierarchical control structure of the MG can overcome the limitation of communication and realize reactive power sharing.In this paper,an improved droop control is adopted,which is based on the hierarchical control structure.The hierarchical control structure consists of zerolevel control,primary control and a proposed secondary control.First,the secondary controller is modeled,and the MG system composed of distributed generators(DGs)is considered as a multi-agent system.The secondary controller can make up for the shortcomings of the droop controller and adjust the frequency and voltage to their rated values.Secondly,the reference voltage and frequency of the zero-level control are calculated,and combined with the primary control.The zero-level control and primary control can make the voltage and frequency of the MG run stably and provide reference voltage for the inverter.Finally,the stability of the system is proved by the theory of multi-agent consistency.A simulation system is established in the Matlab/Simulink environment,and the results show the effectiveness of the proposed controller.

Towards a Service-Oriented Architecture for a Mobile Assistive System with Real-time Environmental Sensing

With the growing aging population, age-related diseases have increased considerably over the years.In response to these, Ambient Assistive Living（AAL） systems are being developed and are continually evolving to enrich and support independent living. While most researchers investigate robust Activity Recognition（AR）techniques, this paper focuses on some of the architectural challenges of the AAL systems. This work proposes a system architecture that fuses varying software design patterns and integrates readily available hardware devices to create Wireless Sensor Networks（WSNs） for real-time applications. The system architecture brings together the Service-Oriented Architecture（SOA）, semantic web technologies, and other methods to address some of the shortcomings of the preceding system implementations using off-the-shelf and open source components. In order to validate the proposed architecture, a prototype is developed and tested positively to recognize basic user activities in real time. The system provides a base that can be further extended in many areas of AAL systems,including composite AR.

Why Hong Kong Ranks Highest in Life Expectancy: Looking for Answers from Data Science and Social Sciences

In trying to explain why Hong Kong of China ranks highest in life expectancy in the world,we review what various experts are hypothesizing,and how data science methods may be used to provide more evidence-based conclusions.While more data become available,we find some data analysis studies were too simplistic,while others too overwhelming in answering this challenging question.We find the approach that analyzes life expectancy related data(mortality causes and rate for different cohorts)inspiring,and use this approach to study a carefully selected set of targets for comparison.In discussing the factors that matter,we argue that it is more reasonable to try to identify a set of factors that together explain the phenomenon.