Search

Convolutiona neural network (CNN) is one of the best neural networks for classification, segmentation, natural language processing (NLP), and video processing. The CNN consists of multiple layers or structural parameters. The architecture of CNN can be divided into three sections: convolution layers, pooling layers, and fully connected layers. The application of CNN became most demanding due to its ability to learn features from images automatically, involving massive amount of training data and high computational resources like GPUs. Due to the availability of the above-stated resources, multiple CNN architectures have been reported. This study focuses on the working of convolution, pooling, and the fully connected layers of CNN architecture, origin of architectures, limitation, benefits of reported architectures, and comparative analysis of contemporary architecture concerning the number of parameters, architectural depth, and significant contribution.

In applied software engineering, the algorithms for selecting the appropriate test cases are used to perform regression testing. The key objective of this activity is to make sure that modification in the system under test (SUT) has no impact on the overall functioning of the updated software. It is concluded from the literature that the efficacy of the test case selection solely depends on the following metrics, namely, the execution cost of the test case, the lines of the code covered in unit time also known as the code coverage, the ability to capture the potential faults, and the code modifications. Furthermore, it is also observed that the approaches for the regression testing developed so far generated results by focusing on one or two parameters. In this paper, our key objectives are twofold: one is to explore the importance of the role of each metric in detail. The secondary objective is to study the combined effect of these metrics in test case selection task that is capable of achieving more than one objective. In this paper, a detailed and comprehensive review of the work related to regression testing is provided in a very distinct and principled way. This survey will be useful for the researchers contributing to the field of regression testing. It is noteworthy that our systematic literature review (SLR) included the noteworthy work published from 2007 to 2020. Our study observed that about 52 relevant studies focused on all of the four metrics to perform their respective tasks. The results also revealed that about 30% of the different categories of regression test case reported the results using metaheuristic regression test selection (RTS). Similarly, about 31% of the literature reported results using the generic regression test case selection techniques. Most of the researchers focus on the datasets, namely, Software-Artefact Infrastructure Repository (SIR), JodaTime, TreeDataStructure, and Apache Software Foundation. For validation purpose, following parameters were focused, namely, the inclusiveness, precision, recall, and retest-all.

The Virtual Kiosk: Taman Herba is a virtual reality application that enables user to explore Taman Herba at UTHM in a virtual world by controlling the movement using keyboard. However, the exposure of benefits of herbs is very low due to no clear sign and there is not enough information about the herbs. Therefore, this application provide user with a virtual world of Taman Herba and the guidance to use the herbs. The virtual world of Taman Herba also includes with the information of each herbs. To ensure this application is well-developed, TRES-D methodology is used during development. This application will be implemented as a kiosk that will be located nearby the garden. This application expected to be a medium to educate committees to preserve our nature and educate them about the benefits of the herbs. After development of this application, most of the users agree that the overall function of this application is good.

Plant is exposed to many attacks from various micro-organism, bacterial disease and pests. The symptoms of the attacks are usually distinguished through the leaves, stem or fruit inspection. Disease that are commonly attack plants are Powdery Mildew and Leaf Blight and it may cause severe damaged if not controlled in early stages. Image processing has widely being used for identification, detection, grading and quality inspection in the agriculture field. Detection and identification disease of a plant is very important especially, in producing a high-quality fruit. Leaves of a plant can be used to determine the health status of that plant. The objective of this work is to develop a system that capable to detect and identify the type of disease based on Blobs Detection and Statistical Analysis. A total 45 sample leaves images from different colour and type were used and the accuracy is analysed. The Blobs Detection technique are used to detect the healthiness of plant leaves. While Statistical Analysis is used by calculating the Standard Deviation and Mean value to identify the type disease. Result is compared with manual inspection and it is found that the system has 86% in accuracy compared to manual detection process.

Improving the quality and quantity of paddy production is very important since rice is the most consumed staple food for billion people around the world. Early detection of the paddy diseases and pests at different stages of growth is very crucial in paddy production. However, the current manual method in detecting and classifying the paddy diseases and pests requires a very knowledgeable farmer and time consuming. Thus, this study attempts to utilize an effective image processing and machine learning technique to detect and classify the paddy diseases and pests more accurately and less time processing. To accomplish this study, 3355 images comprises of 4 classes paddy images which are healthy, brown spot, leaf blast, and hispa was used. Then the proposed five layers of CNN technique is used to classify the images. The result shows that the proposed CNN technique is outperform and achieved the accuracy rate up to 93% as compared to other state-of-art comparative models.

In this paper, a violence video classification model based on deep neural network (DNN) with chicken swarm optimization (CSO) is proposed. Violence is a self-sufficient attribute, the contents that one would not be favorable to see in movies or web videos. This is a challenging problem due to strong content variations among the positive instances of violence. Currently, deep neural network has shown its efficiency in various field that relevant to its implementation and attracts plenty of researchers in awe. However, conservative deep neural network has limitations and a tendency to easily fall into local minima. Regardless of the conventional methods applied to overcome this issue, but these techniques seem insufficiently accurate and does not adopt well to certain webs or user needs. Therefore, the purpose of this study is to assess the classification performances on violence video using Deep Neural Network with Chicken Swarm Optimization (DNNCSO). Hence, in this paper different architectures of hidden layers in DNN have been implemented using the try-error method and the importance of the parameters to examine the effect of the number of hidden layers to the classification performance. The algorithm is evaluated based on error convergence and accuracy. The results have proved the effectiveness of the proposed method up to 77–79% as compared to the conventional DNN which holds 63%. Based on the promising outcome of proposed method, in future the study intended to work on improvised bio-inspired algorithm possibly with different domains and relevant features.

Clustering is the process of breaking down a huge dataset into smaller groups. It has been used in some field studies including pattern recognition, segmentation, and statistics with remarkable success. Clustering is a technique for dividing multivariate datasets into groups. No inherent distance measure on data category makes clustering data more challenging than numerical data. Data category can be assumed following the data from a multinomial distribution. Thus, the standard model parametric model can be used in latent class clustering based on the independent product of multinomial distributions. Meanwhile, multi-valued attributes on the categorical data can be decomposed into the standard set on a multi soft set. In this paper, a clustering technique based on soft set theory is proposed for categorical data through a multinomial distribution. The data will be represented as a multi soft set which is every soft set has its probability of being a member of the cluster. The data with the highest probability will be assigned as the member of the cluster. The experiment of the proposed technique is evaluated based on the Dunn index with regard to the number of clusters and response time. The experiment results show that the proposed technique has the lowest response time with high stability compared to baseline techniques. This study recommends a maximum number of clusters in implementation on the real data.

Every development activity is always related to human or community aspects. This can also lead to changes in the characteristics of the community. The community's increasing awareness and critical attitude need to be accommodated to avoid the emergence of social conflicts in the future. This research is to find out how the public perception about the impact of development on the environment. Two methods are used, i.e., MDA (Maximum Dependency Attribute) and MSMD (the Multi soft set multivariate distribution function). The MDA is to determine the most influential attribute and the Multi soft set multivariate distribution function (MSMD) is to group the selected data into classes with similar characteristics. This will help the police producer plan the right mediation and take quick activity to make strides in the quality of the social environment. The experiment conducted level of impact based on the clustering results with the greatest number of member clusters is cluster 1 (very low impact) with 32.25 % of total data following cluster 5 (Very High impact) with 24.25 % of total data. The experiment obtains the level of impact based on the clustering results. The greatest number of member clusters is cluster 1 (extremely low impact) with 32.25 % of total data following cluster 5 (Very High impact) with 24.25 % of total data. The scatter area impact is spread at districts 6, 7, 10, 11, the most of very high impact and districts 1,2,3,4,5,8 the lowest impact.

Dimension reduction approach is one of the main data reduction approaches in order to reduce the storage and processing time while maintaining the integrity of the original data. A wide range of dimension reduction approaches are based on classical approaches such as PCA and Bayer’s, and machine learning approaches such as clustering, and feature selection techniques. However, many of the approaches do not consider the incomplete information systems where some attribute values are missing or incomplete. Only few studies were proposed for the problem in incomplete information systems due to its complexities, specifically on attribute selection. The most popular approaches is based on probability theory to replace missing values with the most common values, or remove the missing objects from the information systems. However, it needs to know the probability distribution of data in advance. To overcome these issues, we propose a new approach based on conditional entropy to reduce dimensionality. The results show that the proposed approach achieves better data reduction with higher accuracy for objects and dimensionality reduction in incomplete information systems.

Violence is self-sufficient, it is perplexing due to visibility of content dissimilarities among the positive instances that been displayed on media. Besides, the ever-increasing demand on internet, with various types of videos and genres, causes difficulty for a proper search of these videos to ensure the contents is humongous. It involves in aiding users to choose movies or web videos suitable for audience, in terms of classifying violence content. Nevertheless, this is a cumbersome job since the definition of violence is broad and subjective. Detecting such nuances from videos becomes technical without a human’s supervision that can lead to conceptual problem. Generally, violence classification is performed based on text, audio, and visual features; to be precise, it is more relevant to use of audio and visual base. However, from this perspective, deep neural network is the current build-up in machine learning approach to solve classification problems. In this research, audio and visual features are learned by the deep neural network for more specific violence content classification. This study has explored the implementation of deep neural network with monarch butterfly optimization (DNNMBO) to effectively perform the classification of the violence content in web videos. Hence, the experiments are conducted using YouTube videos from VSD2014 dataset that are publicly available by Technicolor group. The results are compared with similar modified approaches such as DNNPSO and the original DNN. The outcome has shown 94% of violence classification rate by DNNMBO.

This paper investigates the existing practices and prospects of medical data classification based on data mining techniques. It highlights major advanced classification approaches used to enhance classification accuracy. Past research has provided literature on medical data classification using data mining techniques. From extensive literature analysis, it is found that data mining techniques are very effective for the task of classification. This paper analysed comparatively the current advancement in the classification of medical data. The findings of the study showed that the existing classification of medical data can be improved further. Nonetheless, there should be more research to ascertain and lessen the ambiguities for classification to gain better precision.

This paper compares between four instructional design approaches used to enhance pupils’ numeracy skill. A literature review performed on four approaches that have potential in enhancing pupil’s numeracy skill. Those approaches are e-Learning, Mobile Learning, Gamification and Problem Based Learning (PBL). The review focuses on design and development of the selected approaches and compares those approaches based on previous studies. The literature survey suggests that e-Learning and gamification are the most suitable instructional design tools to enhance pupils’ numeracy skill.

Digital mammograms are coupled with noise which makes de-noising a challenging problem. In the literature, few wavelets like daubechies db3 and haar have been used for de-noising medical images. However, wavelet filters such as sym8, daubechies db4 and coif1 at certain level of soft and hard threshold have not been taken into account for mammogram images. Therefore, in this study five wavelet filters namely: haar, sym8, daubechies db3, db4 and coif1 at certain level of soft and hard threshold have been considered. Later, peak signal to noise ratio and mean squared error values are calculated. From the obtained results, it can be concluded that db3 (46.44656 db for hard threshold and 43.80779 db for soft threshold) is more appropriate filter for de-noising mammogram images while compared with other wavelets filters.

Nowadays, there are various types of learning materials used in the process of teaching and learning of Al-Quran including the use of mobile application. However, the features of mobile application that are appropriate for the process of memorizing the Al-Quran, especially for the needs of children with autism is still limited. Thus, this paper proposes an interactive Al-Quran mobile application namely iHafaz to facilitate autism children recite and memorizing Al-Quran. A takrir (repetition) technique in Islamic learning approach is embedded in this mobile application in order to assist autism children memorizing the Al-Quran easily. This mobile application consists of two main modules which are Hafaz (Memorize) and Latihan (Exercise). Result from the user testing shows that 72.4% of respondents agree that the takrir technique embedded in the mobile application able to improve the usability of the mobile application in helping the autism children to recite and memorize the Al-Quran easily.

We have proposed a method to assist children with coordination difficulties or dyspraxia to improve their handwriting skills. We have chosen an animation technique called ‘Rotoscopy’, a method that normally been used in animation and film production and adapted it to Rotoscopy Prehandwriting Interface (RPI) prototypes using the interactive whiteboard (IWB) as interaction device. The motivation of this research is to discover how efficient if Rotoscopy is used beyond its normal purposes? Does it give benefits in terms of behavioural and motivational aspect rather than commercial and profit point of view? Implementation of RPI prototypes has taken place through series of workshops with a teacher and a group of children with handwriting difficulties at a special education school in Caerphilly, Cardiff, United Kingdom. In the workshops children were given pre-handwriting activities in two different environments. They have been trained to use RPI prototypes and IWB as well as using pen and paper. Their activities and action has been observed and recorded using video camera. Evaluation method is based-on video analysis of children’s pre-handwriting result and their reaction and motivation during the workshop. It was learnt that majority of children who used RPI prototypes and IWB have produced better results in terms of accuracy of the drawing as compared to results of pen and paper activities. Furthermore the children are more motivated to use the prototypes and IWB rather than using pen and paper. The study’s contribution includes offering a new way to improve children’s prehandwriting skills using computer animation technique and touch-based devices.

Sabah is popular with its beautiful vacation gateway. However there is no game application that related to the interesting places in Sabah as a way for to attract school children about interesting places in Sabah. Therefore, Ally’s Adventure in Sabah which is a 2D mobile game application has been developed to solve this problem. Ally’s Adventure application consist a total 4 levels that developed based on 4 different popular places in Sabah and it have been implemented in Android platform. Game Development Life Cycle (GDLC) has been used as the methodology in the application development process. The beta version of Ally’s Adventure had been tested by its target user; 10 to 15 years old in order to achieved the objectives of its development purposes. It received a good amount of positive feedback and few improvements had been done to the beta version of Ally’s Adventure in Sabah.

Paper presented at AIP Conference Proceedings 1891, 020112 (2017), available at: https://doi.org/10.1063/1.5005445.

Recent advances in the field of image processing have revealed that the level of noise in mammogram images highly affect the images quality and classification performance of the classifiers. Whilst, numerous data mining techniques have been developed to achieve high efficiency and effectiveness for computer aided diagnosis systems. However, fuzzy soft set theory has been merely experimented for medical images. Thus, this study proposed a classifier based on fuzzy soft set with embedding wavelet de-noising filters. Therefore, the proposed methodology involved five steps namely: MIAS dataset, wavelet de-noising filters hard and soft threshold, region of interest identification, feature extraction and classification. Therefore, the feasibility of fuzzy soft set for classification of mammograms images has been scrutinized. Experimental results show that proposed classifier FussCyier provides the classification performance with Daub3 (Level 1) with accuracy 75.64% (hard threshold), precision 46.11%, recall 84.67%, F-Micro 60%. Thus, the results provide an alternative technique to categorize mammogram images.

Content-based Image Retrieval (CBIR) aids radiologist to identify similar medical images in recalling previous cases during diagnosis. Although several algorithms have been introduced to extract the content of the medical images, the process is still a challenge due to the nature of the feature itself where most of them are extracted in low level form. In addition to the dimensionality reduction problem caused by the low-level features, current features are also insufficient to convey the semantic meaning of the images. This paper reviews the recent works in CBIR that attempts to reduce the semantic gap in extracting the features from medical images, precisely for mammogram images. Approaches such as the use of relevance feedback, ontology as well as machine learning algorithms are summarized and discussed.

Effective use of feature set and selection of a suitable classification method are significant for improving classification accuracy. However, mammogram images classification is affected by many factors such as additive gaussian noise, low contrast and artifacts. Therefore, the aim of this paper is to observe the impact of presence /absence of noise on the quality and classification accuracy of mammogram images. The proposed methodology involved five steps that are data collection, images de-noising using wavelet hard and soft thresholding, region of interest (ROI) identification, feature extraction (statistical texture features), and classification. Hundred and twelve images (68 benign images and 51 malignant images) were used for experimental set ups. The experimental results show the improvement of classification accuracy in the presence of noise using wavelet filter with fuzzy soft set classifier compared with the results in the absence of noise within existing classification algorithms.

The noisy nature of digital mammograms and low contrast of suspicious areas which make medical images de-noising a challenging problem. Therefore, image de-noising is an important task in image processing, thus the use of wavelet transform provides better and improved quality of an image and reduces noise level. For medical images, many wavelets like db1, sym8, coif1, coif3 etc can be used for de-noising of a medical image. However, in this paper, haar, sym8 daubechies db3 (mallat), daubechies db4 at certain level of soft and hard threshold have been calculated. Later, peak signal to noise ratio (PSNR) values are calculated for these wavelet methods. These experiments help to select the best wavelet transform for the de-noising of particular medical images such as mammogram images.

Basically, data pre-processing is an important part of data mining. Normalization is a pre-processing stage for any type of problem statement, especially in video classification. Challenging problems that arises in video classification is because of the heterogeneous content, large variations in video quality and complex semantic meanings of the concepts involved. Therefore, to regularize this problem, it is thoughtful to ensure normalization or basically involvement of thorough pre-processing stage aids the robustness of classification performance. This process is to scale all the numeric variables into certain range to make it more meaningful for further phases in available data mining techniques. Thus, this paper attempts to examine the effect of 2 normalization techniques namely Min-max normalization and Z-score in violence video classifications towards the performance of classification rate using Multi-layer perceptron (MLP) classifier. Using Min-Max Normalization range of [0,1] the result shows almost 98% of accuracy, meanwhile Min-Max Normalization range of [-1,1] accuracy is 59% and for Z-score the accuracy is 50%.

Musical instrument classification has a great importance in data mining and multimedia. The number of studies investigating classification of musical instruments using sophisticated modeling such as neural networks, support vector machines, decision tress and rough set. However, the viability of soft set theory for musical instruments classification has not been widely experimented. Thus, this paper introduces a classification system which uses notation of soft set theory incorporating non-western musical instruments i.e. Traditional Pakistani Musical Instruments. One of the factors that may contribute to this phenomenon which are audio length, frame size and starting point of files have been investigated that might affect the performance of the classification algorithm. The modeling process comprises of three steps which are data-preprocessing, dataset partitioning and classification. Experimental results show that 94.26% was obtained from the generated datasets. Soft set theory provides fruitful investigation for musical instruments classification. These results have further expanded the scope of soft set theory for decision making applications.

This paper presents an alternative feature ranking technique for Traditional Malay musical instruments sounds dataset using rough-set theory based on the maximum degree of dependency of attributes. The modeling process comprises seven phases: data acquisition, sound editing, data representation, feature extraction, data discretization, data cleansing, and finally feature ranking using the proposed technique. The results show that the selected features generated from the proposed technique able to reduce the complexity process.

Feature selection has become very important research in musical instruments sounds for handling the problem of ‘curse of dimensionality’. In literature, various feature selection techniques have been applied in this domain focusing on Western musical instruments sounds. However, study on feature selection using rough sets of non-Western musical instruments sounds including Malay Traditional musical instruments is inadequate and still needs an intensive research. Thus, in this paper, an alternative feature selection technique using rough set theory based on Maximum Degree of dependency of Attributes (MDA) technique proposed by [8] for Traditional Malay musical instruments sounds is proposed. The modeling process comprises eight phases: data acquisition, sound editing, data representation, feature extraction, data discretization, data cleansing, feature selection using proposed technique and feature validation via classification. The results show that the highest classification accuracy of 99.82% was achieved from the best 17 features with 1-NN classifier.

In presenting the appropriate data sets, various data representation and feature extraction methods have been discovered previously. However, almost all the existing methods are utilized based on the Western musical instruments. In this study, the data representation and feature extraction methods are applied towards Traditional Malay musical instruments sounds classification. The impact of five factors that might affecting the classification accuracy which are the audio length, segmented frame size, starting point, data distribution and data fraction (for training and testing) are investigated. The perception-based and MFCC features schemes with total of 37 features was used. While, Multi-Layered Perceptrons classifier is employed to evaluate the modified data sets in terms of the classification performance. The results show that the highest accuracy of 97.37% was obtained from the best data sets with the combination of full features.

In recent year, most studies concerned with the classification of musical instruments sounds focus on western musical instruments. With the enormous amount of instruments data and features schemes, adapting the existing techniques for classifying the traditional Malay musical instruments sounds might not be as easy due to the differences in the sounds samples used. Thus, the existing framework and techniques that have been proposed for automatic musical instruments sounds classification system will be reviewed and evaluated especially on their performance in achieving the highest accuracy rate. As a result, a new framework for Traditional Malay Musical Instruments Sounds Classification System and the classification accuracy achieved in the preliminary experiment are presented.