Diabetic Retinopathy Detection System from Retinal Images
Aditi D. Lotliker and Asst. Prof. Amit Patil, Department of Computer Engineering, Goa College of Engineering, Farmagudi-Ponda, India
Diabetes Mellitus is a disorder in metabolism of carbohydrates, and due to lack of the pancreatic hormone insulin sugars in the body are not oxidized to produce energy. Diabetic Retinopathy is a disorder of the retina resulting in impairment or vision loss. Improper blood sugar control is the main cause of diabetic retinopathy. That is the reason why early detection of retinopathy is crucial to prevent vision loss. Appearance of exudates, microaneurysms and hemorrhages are the early indications. In this study, we propose an algorithm for detection and classification of diabetic retinopathy. The proposed algorithm is based on the combination of various image-processing techniques, which includes Contrast Limited Adaptive Histogram Equalization, Green channelization, Filtering and Thresholding. The objective measurements such as homogeneity, entropy, contrast, energy, dissimilarity, asm, correlation, mean and standard deviation are computed from processed images. These measurements are finally fed to Support Vector Machine and k-Nearest Neighbors classifiers for classification and their results were analyzed and compared.
Diabetic Retinopathy, Retinal Images, Support Vector Machine, k—Nearest Neighbors.
Bio-Inspired Optimization Algorithms in Data Mining
Arwa Alrawais, College of Computer Engineering and Sciences, Prince Sattam Bin Abdulaziz University, Al-Kharj 11942, Saudi Arabia
Bio-inspired algorithms are group of algorithms designed based on the observed of natural performs. These inspiration techniques are modelled and used to solve several optimization problems in data mining. This work presents an overview of the popularity of bio-inspired algorithms and its applications. Furthermore, it analyses the bio-inspired algorithms in terms of algorithm behaviour, advantages and potential solution objectives.
Bio-inspired algorithms, data mining algorithms, data mining applications.
Validation Method to Improve Behavioral Flows on UML Requirements Analysis Model by Cross-checking with State Transition Model
Hikaru Morita and Saeko Matsuura, Graduate School of Engineering and Science, Shibaura Institute of Technology, Minuma-ku 307, Saitama, Japan
We propose a method to evaluate and improve the validity of required specifications by comparing models from different viewpoints. Inconsistencies are automatically extracted from the model in which the analyst defines the service procedure based on the initial requirement; thereafter, the analyst automatically compares it with a state transition model from the same initial requirement that has been created by an evaluator who is different from the analyst. The identified inconsistencies are reported to the analyst to enable the improvement of the required specifications. We develop a tool for extraction and comparison and then discuss its effectiveness by applying the method to a requirements specification example.
Requirements Specification, UML Modeling, Validation, Behavior Model.
The Approach of Open Source License Choice Based on Software Dependencies
Zhiyou Liu, Zili Zhang, Zhiqiang Wang and Jing Peng, College of Software, Computer and Information science, Southwestern University, Chongqing Beibei 400715, China
Usually, developers choose an appropriate open source license by looking over the license list of Open Source Initiative or community recommendation. However, the study found that, there are a varieties of open source licenses and software in large numbers developers have difficulty understanding the terms of open source licenses. Therefore, when choosing a license, it is easy to ignore the compatibility risk of open source licenses and fail to protect the legitimate rights and interests of software projects. In response to these issues, we conducted extensive research on factors affecting the choice of open source license. It demonstrate that developers should consider the factor of software dependencies and refer to similar open source projects. For this reason, an open source license choice approach based on software dependencies is proposed. First, the open source projects dependencies data is used to building a dependencies network to computer similarities of projects. Next, a license compatibility detection process of open source projects is implemented in the rule graph of license compatibilities, which is based on data of open source projects dependencies. Finally, provided user with an important reference for choosing an appropriate licenses. The results show that the licenses for 86.207%（Success Rate） of the test projects can be matched and offer the possibility to detect license compatibility in software projects by using the approach.
Open Source License, Open Source Project, Compatibility Risk, Software Dependencies.
Intelligent system for solving problems of veterinary medicine on the example of dairy farms
Shopagulov Olzhas, Tretyakov Igor and Ismailova Aisulu, Dept. of Information Systems, Kazakh Agro, Technical University named after S. Seifullin, Kazakhstan
This article describes an automated expert system developed to diagnose cow diseases and assist veterinarians in treatment. We set before a diagnostic method based on the analysis of observed symptoms and experience of veterinarians. The system represents a web interface for maintaining a database of diseases, their symptoms and treatment methods, as well as a smartphone application for the diagnostics in offline mode. The developed intelligent system will allow agricultural producers to make specific decisions based on automated data analysis. Also presented in the article the information on the developed expert system, and the results of tests and testing during its use. The economic efficiency and importance of the work is determined by the possibility of automated recording of data on the livestock of animals, zootechnical and veterinary operations.
Intelligent system, diagnosis of diseases, application evaluation, milk yield, herd management.
LASTD: A Manually Annotated and Tested Large Arabic Sentiment Tweets Dataset
Kariman Elshakankery, Magda B. Fayek and Mona F. Ahmed, Department Computer Engineering, Faculty of Engineering, Cairo University, Egypt
With the growing attention towards Arabic Sentiment Analysis (SA), the availability of annotated dataset has raised. Although acquiring dataset from social media platforms, microblogs and so on is an easy task, annotation is the hard part. Dataset annotation requires a lot of manual tedious work which stands as a major problem. In addition to that, some datasets are built in house and aren’t available for public access. This paper introduces the LASTD which is a manually annotated dataset for Arabic tweets sentiment analysis along with an insight of its statistics and benchmarks. It consists of more than 15K Arabic tweets annotated as positive, negative and neutral.Using 10-cross validation, three different classifiers were trained and tested for 3-class classification problem and 2-class classification problem. The support vector machine (SVM) classifier tends to have the highest accuracy.LASTD is made public for academic research.
Annotated Dataset, Arabic Tweets Sentiment Analysis, Hybrid Approach, Opinion Mining, Sentiment Analysis, Sentiment Classification.
Genetic Switches Between Two Population with Regards To MRNA and Proteins Applying Markov Chain Stochastic Model Check
Qin He, RubinWang, XiaochuanPan, East China University of Science and Technology, Meilong 130
Arc, one virus-like gene, crucial for learning and memory, was dis-covered by researchers in neurological disorders fields, Arc mRNA’s single directed path and allowing protein binding regional restric-tively is a potential investigation on helping shuttle toxic proteins responsible for some diseases related to memory deficiency. To study especially the transform between mRNA and proteins, the switching function of the phenotypes, ’normals’ multiplying populations and ’persisters’, resilient to stress instead of multiplying is of our interest. Mean time to switching (MTS) is calculated explicitly quantifying the switching process in statistical methods combining Hamiltonian Markov Chain(HMC). The model derived from predator and prey with typeII functional response studies the mechanism of normals with intrin-sic rate of increase and the persisters with the instantaneous discovery rate and converting coefficients. During solving the results, since the numeric method is applied for the 2D approximation of Hamiltonion with intrinsic noise induced switching combining geometric minimum action method. In the application of Hamiltonian Markov Chain, the behavior of the convertion (between mRNA and proteins through 6 states from off to on ) is described with probabilistic conditional logic formula and the final concentration is computed with both Continuous and Discret Time Markov Chain(CTMC/DTMC) through Embedding and Switching Diffusion.The MTS, trajectories and Hamiltonian dynamics demonstrate the practical and robust advantages of our model on interpreting the switching process of genes (IGFs, Hax Arcs and etc.) with respects to memory deficiency in aging process which can be useful in further drug efficiency test and disease curing.
switching model, mean time to switching, Hamiltonian Markov Chain, geometric minimum action method.
Trust Based Routing Protocols in Manet
K.Divya, Ph.D Research Scholar, Department of Computer Science, Gobi Arts & Science College, Gobichettipalayam, INDIA, Dr.B.Srinivasan, Associate Professor, Gobi Arts & Science College, Gobichettipalayam, INDIA
A mobile ad hoc network is a wireless network in which no infrastructure is available. MANET is a self-configuring network. Due to dynamic nature of MANET it is very challenging work to employ a secure route. The intermediate nodes cooperate with each other as there is no such base station or access point. The routing protocols play important role in transferring data. Cryptographic mechanisms are used in routing protocols to secure data packets while transmitted in the network. But cryptographic techniques incur a high computational cost and can’t identify the nodes with malicious intention. So, employing cryptographic techniques in MANET are quite impractical as MANETs have limited resource and vulnerable to several security attacks. Trust mechanism is used as an alternative to cryptographic technique.
Network Protocols, Wireless Network, Mobile Network, Virus, Worms &Trojon.
Secure Critical Infrastructure for National Population Census using Blockchain with Decentralized Business Model
Sana Rasheed and Dr. Soulla Louca, School of Business, University of Nicosia, Cyprus
National population census provides the basis for governments’ financial, economic, health and education policies for its populace. It plays a vital role in mapping country’s growth and financial trajectories and it is the single most valuable and shared resource among government departments and apparatuses. It is generally non-classified, public and has multiple stakeholders. There are few drawbacks with current system by design – database in inherently non-secure due to its public usage by multiple stakeholders, corruption in data is usual due to political and racial gains, data update of individual is a laborious and repeatable process, and there is a dire need to protect the rights of minorities and marginalized communities against corrupt government. Immutability of records is required and appreciated, thus making it an ideal candidate for the blockchain based decentralized solution. This paper presents a secure critical infrastructure for national population census using blockchain with decentralized business model.
Secure critical infrastructure, population census, survey, technology, government transparency, minority rights, information corruption, data protection and transparency.
Lossless Steganography on Orthogonal Vector for 3D H.264 with Limited Distortion Diffusion
Juan Zhao and Zhitang Li, School of Mathematics & Computer Science, Wuhan Polytechnic University, Wuhan430048, China Institute of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan 430074, China
In order to improve the undetectability, a lossless algorithm based on orthogonal vectors with limited distortion diffusion for 3D H.264 video is proposed in this paper. Inter-view distortion drift is avoided by embedding data into frames, which do not predict other views. Three conditions and pairs of coefficients are proposed to prevent intra-frame distortion diffusion. Several quantized discrete cosine transform coefficients are chosen from an embeddable luminance 4×4 block to construct a carrier vector, which is modified by an offset vector. When the carrier vector and the offset vector are orthogonal or near to be orthogonal, a data bit can be hidden. Experimental results indicate that the method is effective by enhancing peak signal-to-noise ratio with 7.5dB and reducing the Kullback-Leibler divergence with 0.07 at least. More than 1.7×1015 ways could be utilized for constructing the vectors, so it is more difficult for others to steal data.
Lossless Steganography, Reversible Data Hiding, Orthogonal Vector, 3D H.264, Distortion Drift.
Reversible Data Hiding based on Two-Dimensional Histogram Shifting
Juan Zhao and Zhitang Li, School of Mathematics & Computer Science, Wuhan Polytechnic University, Wuhan430048, China, Institute of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan 430074, China
This paper presents a two-dimensional histogram shifting technique for reversible data hiding algorithm. In order to avoid the distortion drift caused by hiding data into stereo H.264 video, we choose arbitrary embeddable blocks from 4×4 quantized discrete cosine transform luminance blocks which will not affect their adjacent blocks. Two coefficients in each embeddable block are chosen as a hiding coefficient pair. The selected coefficient pairs are classified into different sets on the basis of their values. Data could be hidden according to the set which the value of the coefficient pair belongs to. When the value of one coefficient may be changed by adding or subtracting 1, two data bits could be hidden by using the proposed method, whereas only one data bit could be embedded by employing the conventional histogram shifting. Experiments show that this two-dimensional histogram shifting method can be used to improve the hiding performance.
Reversible data hiding, Two-dimensional histogram shifting, H.264, Multi-view coding.