The extraction of feature shows a significant part of iris recognition system. The robustness of recognition accuracy mostly depends on efficient extraction of feature. In the development of an effective recognition system, it is required that the best discriminating feature available in an iris pattern to be properly extracted. This paper applied some selected feature extraction techniques: 1D Log-Gabor Filter (1D LGF), 2-D Gabor-Filter (2D GF), Discrete Cosine Tansform (DCT) and Scale Invariant Feature Transform (SIFT) for extraction of iris features and fusion technique. The CASIA iris image dataset was used to evaluate with evaluation parameters: False Acceptance Rate (FAR), False Rejection Rate (FRR), Error Rate (RA) and Recognition Accuracy (RA). The combined 1D Log-Gabor and 2D Gabor filter approach outperformed other techniques with 92.22% of recognition accuracy, FRR of 0.0186, FAR of 0.1052 and ER of 2.87%.
A Model of Intelligent Recommender System With Explicit Feedback Mechanism for Performance Improvement
Recommender Systems are intelligent applications designed to assist the user in a decision-making process whereby user wants to choose one item amongst the potentially overwhelming set of alternative products or services. This work focused on using users bank statements that explicitly shows inflow and outflow of funds. The dataset used is real and reliable because the use of non-reliable data in a recommender system causes users lack of trust in the system. However, the data collected were anonymized for privacy reasons. The recommender system was developed as a web application using Java programming language. Unlike other recommender systems, the graph-oriented database management system was used. In Google news, 38% of the total views are the result of recommendations; similarly, 60% of the rented movies from Netflix come from recommendations and more than that Amazon sales percentage due to recommendations are 35%. Successful integration of recommendation system by online companies like Amazon, eBay, Flipkart amongst others impelled the research community to avail similar benefits in financial domain to recommend product and services (Lim, 2015). Therefore, recommendation systems are considered an expedient factor in business nowadays. The aim of all recommender systems is to provide recommendation that will be favourably evaluated and accepted by its users. This work provides detailed descriptions of methods employed to proffer solutions to intelligent recommender system with explicit feedback mechanism. The methodology of this research work refers to the research approach adopted by the researcher to tackle the research problem as stated in earlier chapter. Since the efficiency and maintainability of any application is solely dependent on how the designs are prepared, this chapter describes the various processes, methods and procedures used to achieve set objectives and the conceptual structure within which the research was conducted.
The transputer (Transistor Computer) was an innovative computer design of the 1980s from INMOS, a British semiconductor company based in Bristol. The transputer was conceived of as a building block for electronic systems comprising a processor, memory and a communication system. The transputer was unique in that each processor had a built-in simple operating system, memory and four high speed (20 Mbit/s full duplex) bi-directional serial links. The transputer is essentially a computer system on a chip. The links on the transputer allow connection to up to four other transputers or peripherals such as video graphics, floppy and hard disc drives, Ethernet networking and standard RS-232 serial ports. In this paper discusses the original purpose of the transputer, the architectural and the network design. It also lay emphasis on the factors that birth the dead end of the tranputer technology and the restoration project.
As the knowledge of wireless technology keeps growing exponentially in the field of telecommunication, new ideas spring up over time to address and proffer solutions to generational wireless communication issues. In this term paper, the reasons for wireless technology growth was explained, and a detailed information on emerging wireless technology was highlighted. The paper highlights the idealistic of heterogeneous networks, how security and short range within the network can be solved through a suggested solution of hybridizing RF and OW.
Research has over time played a pivotal role in mankind’s quest for knowledge and technological advancement. In all spheres of human existence, research and its further application have over time been able to show the obvious, and yet sometimes hidden unity of science and the philosophical and sociological settings in which everything operates. Essentially, research has helped man to explore once thought of as bizarre phenomena and afforded man the opportunity to draw a fine line between opinions and facts towards gaining maximum benefits from the research’s orientation (Williams, 2007). Too frequently, research is viewed as a formalized process of applying a rigid sequence of steps to the solution to a problem but in actual fact, research in itself entails flexibility in order to maximize scientific methods. This paper explains the concept of literature review in research and how a literature review is done in other to enhance the quality of the research work produced.
This work is motivated by the critical role that sediment yield prediction plays in preventing natural and economic disasters. Methods based on regression techniques have been used to solve the problem but they are generally inadequate in predicting river sediment yield because of the inherent complexity of the problem. This work uses the Adaptive Neuro-Fuzzy Inference System (ANFIS) to solve the problem. The ANFIS model accepts four input data namely temperature, rainfall, water stage and water discharge and gives on output data that represents the sediment yield. The ANFIS model was developed and simulated with MATLAB 7.0 using the Levenberg-Marquardt optimization method and trained with a maximum of 1500 epochs at a learning rate of 0.5. the results obtained was compared with the ones obtained with the Artificial Neural Network (ANN) model and it was found that the ANFIS model performs better than the ANN model.
Sport betting companies and participants can maximize their profit in the sports betting business if they are able to accurately predict the outcome of football matches. This work seeks to develop such a football match prediction system with Manchester United football club as a case study. The developed system is based on an Artificial Neural Network (ANN) model. Scores from previous matches played by Manchester United were used to train and validate the network. The system has prediction accuracies of 73.72% and 113.5% for goals scored by, and against Manchester United respectively. The performance of the model is reasonably good but it can be improved by training the model with more football scores.