Volume 2 Number 2 September 2012
1 |
An Analysis on the Main Reasons for using Social Networking among Indian Youth – A Fuzzy Approach Abstract: “Mailing Occasionally, Chatting Frequently and Social Networking Habitually” Youth of today constitute the first generation growing up with internet. Young people use social networking in ways that are radically different from adults, in that they focus on the expressive rather than the informative use. Further, teenagers use the social networking sites for social purposes rather than for coordinating and making work more efficient. This trend shows the soon-to-be critical importance of digital personal and emotional content. This paper reports an informal study that investigated the use of social networking among Indian youth. The focus is on the ways in which youth use these sites as part of their everyday life, the purposes for which they use the social networking sites and how they handle their personal data using them. The purpose is to gain a deeper insight into the social networking usage among Indian youth. |
2 |
A novel approach for nose-tip detection on 3D face images across pose Abstract: This paper investigates a novel technique for extraction of the nose-tip from three-dimensional face image in any pose, which is needed for nose-tip based face registration. The present technique uses weighted median filter for smoothing. No normalization process is applied and the system correctly detects nose-tips across any pose variations. In our system, at first the range images are thresholded using Otsu’s thresholding algorithm, then filling of holes is done using interpolation method and after that smoothing is done using weighted median filtering mechanism. In the last and final step, nose-tip is detected using maximum intensity algorithm. To evaluate the performance of our approach for nose-tip localization, we have used FRAV3D, GavaDB and Bosphorus database. In case of FRAV3D database out of 542 range images, nose-tips were correctly located for 536 images thus giving 98.70% of good nose-tip localization, in contrast to the method without smoothing which accounted for only 521 face images. In case of GavaDB database, nose-tip was correctly recognized for 421 images out of 549 images thus giving 76.68% of good nose-tip localization in contrast to the method without smoothing which accounted for only 405 face images. In case of Bosphorus database the recognition rate for nose-tips was far better than FRAV3D and GAVADB because the present technique detects nose-tips correctly for 4476 images out of 4935 correctly with smoothing in contrast to the method without smoothing which accounted for only 4333 face images. The overall performance of the system is 90.27% with smoothing whereas the original system gave a performance of only 87.27%.From the results we can conclude that maximum intensity technique has great capabilities for nose-tip detection across variant poses and that Bosphorus database has given a better result than FRAV3D and GAVADB database. |
3 |
A Novel Approach to Simplifying Boolean Functions of Any Number of Variables Abstract: Expressions of Boolean functions in the minimal form would be essential for many needs such as hardware designs. There are several ways such as K-map technique and tabular method of Quine-McCluskey to simplify the Boolean expressions. These currently used techniques have drawbacks such as the limitation on number of variables and dependence on ‘minterms’. Hence, these methods do not adapt programming perfectly. A new method has been introduced in this paper to minimise the Boolean functions without considering the minterms. This method deals with the sum of product (SOP) expressions – it takes the input as SOP and gives the output as SOP. Each product in SOP is encoded in a novel way and represented in a row of a table where the columns correspond to the variables involved in the expressions. The encoded products are taken pair by pair and an appropriate rule of the set of four sound rules is applied to simplify. The encoding is used to select the most promising pairs to apply the rules in a systematic way. This process is continued until no further pair selection is possible. This novel idea makes possible that the expression can contain any number of variables without increasing the complexity of the simplification process though it needs a little more work in encoding and in selecting the most promising products. Also, this idea can be easily programmed as the algorithm is very systematic. The implementation of the algorithm in C# and testing proves that the idea works well efficiently though looks simple. In fact, no similar idea is reported to our knowledge. |
4 |
Forecasting Model for Vegetable Price Using Back Propagation Neural Network Abstract: The Agricultural sector needs more support for its development in developing countries like India. Price prediction helps the farmers and also the Government to make effective decision. Based on the complexity of vegetable price prediction, making use of the characteristics of data mining classification technique like neural networks such as self-adapt, self-study and high fault tolerance, to build up the model of Back-propagation neural network (BPNN) to predict vegetable price. A prediction model was set up by applying the neural network. Taking tomato as an example, the parameters of the model are analyzed through experiment. At the end of the result of Back-propagation neural network shows accuracy percentage of the price prediction. |
5 |
A Weighted Utility Framework for Mining Association Rules using Closed Item sets Abstract: Association rule discovery is used to identify relationship between the items from transaction databases. A traditional Association Rule Mining concentrates on qualitative aspects of attributes (significance, utility) as compared to quantitative attributes (no of appearances in a database). The qualitative approach is used for finding the best item sets. This approach does not yield a company’s profit because the frequency of occurrence of items may be less. In Association Rule Mining the weight is associated with each item set by considering the significance of that item set in profit as well as frequency of occurrences of items in transactions. The name of this association rule mining is called Weighted Utility Association Rule Mining. The main challenge is weighted and utility framework does not hold anti-monotonic property. This framework produces many redundant rules. The proposed framework is used to generate non-redundant rules using a closed frequent item sets. This item sets are not losing any interesting and significant item sets. |
6 |
A New Speed Function for Level set Based Deformable Model for Tumor Segmentation in Medical Images Abstract: Tumor segmentation from medical image data is a challenging task due to the high diversity in appearance of tumor tissue among different cases. In this paper we propose a new level set based deformable model to segment the tumor region. We use the gradient information as well as the regional data analysis to deform the level set. At every iteration step of the deformation, we estimate new velocity forces according to the identified tumor voxels statistical measures, and the healthy tissues information. This method provides a way to segment the objects even when there are weak edges and gaps. Moreover, the deforming contours expand or shrink as necessary so as not to miss the weak edges. Experiments are carried out on real datasets with different tumor shapes, sizes, locations, and internal texture. Our results indicate that the proposed method give promising results over high resolution medical data as well as low resolution images for the high satisfaction of the oncologist at the Cancer Treatment Unit at Jaffna Teaching Hospital. |
7 |
Performance Evaluation of Hybridized Rough Set based Unsupervised Approaches for Gene Selection Abstract: Gene Selection aims to find a subset of highly informative genes from a problem domain which retains high accuracy to represent original genes. Rough Set Theory is adopted in this paper to discover the data dependencies and to reduce the number of genes contained in the dataset using the data alone without requiring additional information about the genes. Selecting genes in unsupervised learning scenarios is a harder problem than supervised gene selection due to the absence of class labels that would guide the search for relevant genes. PSO (Particle Swarm Optimization) is an evolutionary computation technique, which finds global optimum solution in many applications. This paper studies the performance of Unsupervised PSO based Relative Reduct (US-PSO-RR) and Unsupervised PSO based Quick Reduct (US-PSO-QR) approaches by applying it for a set of gene expression datasets to find the harmful genes easily. These two algorithms employs a population of particles existing within a multi-dimensional space and dependency measure that combines the benefits of both PSO and rough sets for better data reduction. The effectiveness of the algorithms is measured by using various clustering accuracy indices. |
8 |
A comparative analysis of genetic based feature selection on heart data Abstract: Feature selection has been an active research area in data mining. The main idea of feature selection is to choose a subset of input variables by eliminating features with little or no predictive information. Feature selection can significantly improve the comprehensibility of the resulting classifier models and often build a model that generalizes better to unseen points. In this paper, genetic algorithm and Compound featuristic genetic algorithm are compared. The comparison on feature selection, reduced attributes produced by genetic algorithm is 6 where reduced attributes produced by the Compound featuristic genetic algorithm is 4. In addition, genetic algorithm and Compound featuristic genetic algorithm are compared under non-fuzzy and fuzzy classifier to obtain the highest accuracy. Fuzzy Decision tree, Fuzzy Naive Bayes, Fuzzy Neural network and Fuzzy K-means are studied under the genetic algorithm and Compound featuristic genetic algorithm. Results exhibit that the Fuzzy K-means classification technique outperforms than other three classification techniques after incorporating fuzzy techniques, also Fuzzy K-means under Compound Featuristic Genetic Algorithm produces the higher accuracy than genetic algorithm. The experiments are carried out on public domain datasets available in UCI machine learning repository heart data set and it is implemented in MATLAB. |
9 |
An Approach to Efficiently Recognize Number Plates from Car Images Abstract: The current scenario is envisaging a tremendous growth in the usage of cars. This increase is demanding automated process for many situations like highway electronic toll collection, automatic parking attendance, petrol station forecourt surveillance, speed limit enforcement, security and customer identification enabling personalized services. An Automatic Number Plate Recognition (ANPR) is a system that can be used in these situations. Motion video or camera still images are used to recognize the car’s number plate characters. This paper presents an efficient approach for Automatic Number Plate Recognition which consists of four phases i.e. Number Plate Localization, Preprocessing, Character Segmentation and Optical Character Recognition. The results have been compared with standard methods at each phase and this proposed method presents better results than the existing ones. |
10 |
Efficient Integrated Coding for Compound Image Compression Abstract- The development of computer and network technologies, image with mixed text, graphics and natural picture are seen everywhere, such as captured screen, web page, scanned electronic documents, slides, posters, compound images and so on. Compound image compression is one of the real-time applications of computer screen image transmission. It is used to reduce the amount of data required to present the digital image and to improve the appearance of an image to a human observer and also to extract quantitative information. To compress a compound image various types of lossy and lossless algorithms are used. This research work deals with the preprocessing, macroblock divisions, transformations, quantization, text block and non-text block analysis, lossy and lossless algorithms are used to compress a compound image to produce a high compression ratio, less compression time and decompression time and high PSNR value than the existing method. |