Networks

We propose an optical 3R regenerator that incorporates a clock enhancement stage based on a semiconductor optical amplifier (SOA)-based Mach-Zehnder interferometer and a 10-GHz all-optical clock recovery module employing a Fabry-Perot filter and an SOA. The experiments assess the optical 3R regeneration technique using a fiber recirculation loop containing 125-km dispersion uncompensated large effective area fiber with a total chromatic dispersion of 531.25 ps/nm. The optical 3R regeneration achieves error-free 125 000-km dispersion uncompensated return-to-zero transmission at 10 Gb/s over 1000 optical 3R stages. The bit-error-rate (BER) measurements show that there is only 1.2-dB power penalty at 10 9 BER between 125 000-km (Lap 1000) uncompensated transmission and back-to-back using pseudorandom bit sequence 2 23 1. Index Terms-All-optical clock recovery, dispersion uncompensated transmission, Fabry-Pérot filter (FPF), optical regeneration, semiconductor optical amplifier-based Mach-Zehnder interferometer (SOA-MZI), synchronous modulation, wavelength conversion.

Using asymmetric narrow-band optical filtering to improve the response and reshape the optical spectrum of a commercial directly modulated distributed feedback laser, we demonstrate 40-Gb/s error-free transmission over 25-km negative dispersion fiber without dispersion compensation.

We experimentally demonstrate a polarization-mode dispersion (PMD) monitoring technique that is insensitive to chromatic dispersion (CD) utilizing a polarizer and a low-speed detector for an 80-Gb/s polarization-multiplexed return-to-zero differential phase-shift-keying (pol-muxed RZ-DPSK) data channel. Measured RF power increment of 16.2 dB, which is insensitive to 0 approximately 100 ps/nm CD, is measured in the presence of the increasing differential group delay (DGD) from 0 to 12 ps. High-speed components are not required for monitoring the PMD on the pol-muxed data channel, which means that the proposed technique is potentially applicable to the higher speed pol-muxed data channels.

This paper presents an approach to recognize facial expressions on human faces. For this research work, facial features, such as the lips, the chin region and especially the eyes, were extracted for dimensionality reduction and a unique face representation using the Principal Component Analysis (PCA). However prior to doing this, the frontal faces were normalized using adaptive histogram equalization for removal of noise and illumination invariance. Experimental results demonstrate that this technique can efficiently classify expressions from the face, with 92% accuracy, using the Support Vector Machine (SVM). The system was implemented using the MATLAB programming language.

Machine learning has been applied to various domains such as pattern recognition, image recognition, fraud detection, medical diagnosis, banking, bioinformatics, commodity trading, computer games and various control applications. Recently, there has been a paradigm shift towards the educational sector in evaluating higher education tasks. The focus of this work is on identifying a suitable algorithm for predicting tertiary students' academic performance, based on their family background factors and previous academic achievements. A university case study with 204 enrolment records of students admitted into computer science program in Kwara State University, Nigeria between 2013 and 2015 was used. The students' first year academic performance was measured by Cumulative Grade Point Average (CGPA), at the end of the first session and the previous academic achievement was measured by SSCE grade score and UME score. The MATLAB programming language was used to carry out the experimental research with emphasis on C4.5 and Randomtree algorithm. These algorithms were compared, using the holdout method, considering, accuracy level, confusion matrices and CPU time to determine the optimal model. Also, a framework of a predictive system based on the rules generated from the optimal model was designed.

One of the most widely used techniques often employed in mining knowledgeable information from medical data bases are Data Mining techniques. While Data mining techniques have proved to be effective in building models illustrating important data classes, especially where class attribute is involved in the construction of the classifier, K-Nearest Neighbor (KNN) is has proved to be an indispensible tool, in pattern recognition. The volume of Medical data bases are huge in nature and getting the right component from them, is essential for data scaling and better data normalization before presenting them for classification. Heart disease is one of the leading causes of death in the world today and based on statistics, the number of adults living with heart failure, increased from about 5.7 million between 2009-2012, to about 6.5 million between 2011-2014 (according to the American Heart Association’s 2017 Heart Disease and Stroke Statistics Update).There is therefore need to provide medical practitioners with decision support systems that can help them in the early detection of heart conditions. In this paper, the Probabilistic Principal Component Analysis (PPCA), is used to preprocess and extract components from a clinical dataset, in order to provide a more organized and detailed data to be passed into the KNN classifier, thereby allowing for better classification. Experimental results shows that the combination of the PPCA and KNN is very productive for the prediction of heart diseases as it achieved an accuracy of 98.08%.

The microarray classification technique is capable of tracking the expression levels of thousands of genes simultaneously. The high dimensional feature vectors of microarray, impose a high dimensional cost, as well as the risk of over fitting during classification. Thus, it is necessary to reduce the dimension, through the use of an efficient method such as; feature selection. In this paper, a different technique, based on classification of micro array cancer dataset, including the two basic approaches of feature selection: the filter and wrapper techniques have been employed.

Artificial intelligence and expert systems have gone a long away in proffering solutions to decision making problem in various sectors. One of the major challenges faced today in under developed country is access to quality and fast health facilities, which poses a big threat to the health a condition of patients. Accurate medical diagnosis is one of the major ways to sustain good health and live long. In this paper, a neuro-expert system was developed using advanced neuro-fuzzy inference system, taking into consideration the combination of eight attributes or factors and one output for the prediction and diagnosis of cervical cancer. Cervical cancer dataset obtained from cancer medical experts, was used to build the system. An evaluation performance was so as to carried out to determine the level of predictive and explanatory power of the developed system. The resulting test carried on the systems shows a very good predictive model with an accuracy of 93.54%.

Understanding customer needs is crucial to gaining and retaining customers in a web store, on online e-commerce applications. In order to ensure a versatile system for e-commerce, the pattern generated by customers when they click icons to select some particular products on an e-commerce website, based on their choice, ought to be studied, recorded and built-up into a database. Data mining techniques can then be applied in mining and analysing information for this database in order to help wholesaler and retailers, improve sales, marketing strategies and product advertisement. In this paper, a conceptualized a framework for determining customers' product choice and factors involved in choosing online commodities is proposed. The system is tailored for phone products.

Encryption is used to transmit data securely on open networks. Information contents may be textual data or image data. Encryption of text or images, which cover the highest portion of multimedia file transfer, is most important, during secure transmission of information. Several techniques have been used to protect confidential image data from unauthorized access. In this paper, the Data Encryption Standard (DES) algorithm was used as an encryption method for iris images, the encryption and decryption speed and the memory used up in the process, being observed. The system, which was implemented using the MATLAB environment, indicates a slower time for encryption but a faster time for decryption using the DES algorithm, thus showing that it is faster to decrypt than to encrypt with the DES cipher key. This guarantees the security of images using the cipher key DES algorithm and thus establishes that the DES algorithm presented assures resilient security for the CASIA iris image.

ABSTRACT: Surface Electromyography (sEMG) signals have been found to be useful in developing methods of recognizing hand gestures using digital signal processing devices or ensemble approaches. In building hand gesture recognition models, it is essential to avoid or reduce computational complications which may result in complex circuit connections while implementing the model. As such, simple but efficient approaches in achieving the desired results are required. A explicable model for hand gesture recognition, using Flexible Neural Trees (FNTs) and sEMG signals is presented in this paper. sEMG is an approach of detecting and documenting the electrical impulses of the muscles, from the surface of the skin. The feed forward Neural Network approach is first used in generating the FNT model before improving it, using predefined simple instruction sets. The FNT model helped to avoid complex computations while building the model and provided a high recognition rate. The experimental outcomes obtained from the developed model shows that the model is capable of classifying six different hand gestures at an accuracy of 98.56%.

In this paper, a content-based image retrieval system (CBIR), which computes HSV color space similarity among images is presented. Content-based image retrieval process consists of computing a feature vector that characterizes the nature of the images. The images are stored in a feature database. In CBIR system, the user provides a query image and the system calculates the feature vector, and then compares it with a certain image features of the image database. This document gives a brief description of a system developed for retrieving images similar to a query image from a large set of distinct images. It follows an image processing based approach to extract features present in an image. These features are stored in vectors called feature vectors and compared to the feature vectors of query image and thus, the image database is sorted in decreasing order of similarity. An image is partitioned into sub-blocks of equal size as a first step. Color of each sub-block is extracted by quantifying the HSV color space into non-equal intervals and the color feature is represented by the 3D histogram. A one-to-many matching method is used to compare the query and target image. Euclidean distance is used in retrieving the similar images. The efficiency of the method is demonstrated with the graphical results.

Typhoid fever is one of the major life threatening diseases, accounting for the death of millions of people every year apart from contributing to economic backwardness, mostly in Africa. Prompt and accurate diagnosis is a major key in the medical field, the large number of deaths associated with typhoid fever is as a result of many factors which include: poor diagnosis, self-medication, shortage of medical experts and insufficient health institutions. These prompted for the development of a typhoid diagnosis system that can be used by anyone of average intelligence as this will assist in quick diagnosis of the disease despite shortage of health institutions and medical experts. A fuzzy logic technique was used on the labeled set of typhoid fever conditional variables to generate explainable rules for the diagnosis of typhoid fever. The labeled database was divided into five different levels of severity of typhoid fever and the classification accuracies on both the training set and testing set are 95% and 96% respectively. Implementation was carried out using Matlab as front end and MySQL as database.

With the huge progress made in data exchange by electronic systems, the need of information security has become a necessity. Due to growth of multimedia application, security has become an important issue of communication and storage of files. Symmetric key cryptography is a common cryptographic technique, which involves, using the same key at both the transmitter and receiver side. The main advantage of symmetric key encryption is its less computational cost, compared to its counterpart-public key encryption. This research work, evaluated the performance of two symmetric cryptography techniques using the Advanced Encryption Standard (AES) and Data Encryption Standard (DES) for the security of plain text messages. The experimental approach was carried out with MATLAB Programming language. The two algorithms scaled on 64 bit key size for its encryption and their performance was measured based on the encryption and decryption time alongside with the encryption and decryption memory. The performance evaluation shows that when it comes to the computational time, the Data Encryption Standard performs faster than the Advanced Encryption Standard and also takes more of a Higher CPU Memory in processing than the AES symmetric key cryptography method.

One of the most widely used techniques often employed in mining knowledgeable information from medical data bases are Data Mining techniques. While Data mining techniques have proved to be effective in building models illustrating important data classes, especially where class attribute is involved in the construction of the classifier, K-Nearest Neighbor (KNN) is has proved to be an indispensible tool, in pattern recognition. The volume of Medical data bases are huge in nature and getting the right component from them, is essential for data scaling and better data normalization before presenting them for classification. Heart disease is one of the leading causes of death in the world today and based on statistics, the number of adults living with heart failure, increased from about 5.7 million between 2009-2012, to about 6.5 million between 2011-2014 (according to the American Heart Association's 2017 Heart Disease and Stroke Statistics Update).There is therefore need to provide medical practitioners with decision support systems that can help them in the early detection of heart conditions. In this paper, the Probabilistic Principal Component Analysis (PPCA), is used to preprocess and extract components from a clinical dataset, in order to provide a more organized and detailed data to be passed into the KNN classifier, thereby allowing for better classification. Experimental results shows that the combination of the PPCA and KNN is very productive for the prediction of heart diseases as it achieved an accuracy of 98.08%.

Artificial intelligence and expert systems have gone a long away in proffering solutions to decision making problem in various sectors. One of the major challenges faced today in under developed country is access to quality and fast health facilities, which poses a big threat to the health a condition of patients. Accurate medical diagnosis is one of the major ways to sustain good health and live long. In this paper, a neuro-expert system was developed using advanced neuro-fuzzy inference system, taking into consideration the combination of eight attributes or factors and one output for the prediction and diagnosis of cervical cancer. Cervical cancer dataset obtained from cancer medical experts, was used to build the system. An evaluation performance was so as to carried out to determine the level of predictive and explanatory power of the developed system. The resulting test carried on the systems shows a very good predictive model with an accuracy of 93.54%.

Understanding customer needs is crucial to gaining and retaining customers in a web store, on online e-commerce applications. In order to ensure a versatile system for e-commerce, the pattern generated by customers when they click icons to select some particular products on an e-commerce website, based on their choice, ought to be studied, recorded and built-up into a database. Data mining techniques can then be applied in mining and analyzing information for this database in order to help wholesaler and retailers, improve sales, marketing strategies and product advertisement. In this paper, a conceptualized a framework for determining customers’ product choice and factors involved in choosing online commodities is proposed. The system is tailored for phone products.

A report from the International Institute of Tropical Agriculture (IITA) shows a record of over 50 improved cassava varieties that are currently developed to improve the quality and quantity of the industrial outputs derived from cassava. These varieties have been in circulation to both small and large scale farmers since 2013. Over the time, it has been discovered from an IITA survey, that farmers have challenges in deciding the cassava variety that would provide the highest percentage yield of desired industrial outputs (Ethanol, flour, Garri, Starch). The trial and error methods adopted to determine the variety to plant has led to wastage, and inconclusive results. Using various methods to guide their decision can be very cumbersome. There are difficulties in tracing track records, as well as retrieval of certain data set from bunch of records. Loss of record or data set would lead to starting the whole methods all over, which is not an efficient way of solving the problem. Objective: This study designed and implemented a decision support system for cassava farming in Nigeria. Method: This study adopts PHP, JavaScript, and HTML at the front-end of a user friendly application developed to handle large records-keeping processes, with MYSQL database at its backend. Results: The DSS produced 3376kg while manual method produced 3200kg of Starch, the DSS yielded 125kg while manual method produced 0kg of Protein and the DSS generated 1500kg while manual method generated 1450kg of Garri when 5000kg of fresh yield of Cassava variety TMS 4(2)1425 was considered. The DSS indicated highest records of farm outputs over manual method. Conclusions: Digital literacy is not imperative for its operation, thereby proffering solutions to the challenge of informed variety investment decisions in small/large cassava farming activities. Against the wastage from previous methods, experimental output from this developed system indicates higher percentage yield of desired industrial cassava output.