Search

Federated learning (FL), as a paradigm for addressing challenges of machine learning (ML) to be applied in private distributed data provides a novel and promising scheme to promote ML in multiple independently distributed healthcare institutions. However, the non-IID and unbalanced nature of the data distribution can decrease its performance, even resulting in the institutions losing motivation to participate in its training. This paper explored the problem with an in-hospital mortality prediction task under an actual multi-center ICU electronic health record database that preserves the original non-IID and unbalanced data distribution. It first analyzed the reason for the performance degradation of baseline FL under this data scenario, and then proposed a personalized FL (PFL) approach named POLA to tackle the problem. POLA is a personalized one-shot and two-step FL method capable of generating high-performance personalized models for each independent participant. The proposed method, POLA was compared with two other PFL methods in experiments, and the results indicate that it not only effectively improves the prediction performance of FL but also significantly reduces the communication rounds. Moreover, its generality and extensibility also make it potential to be extended to other similar cross-silo FL application scenarios.

The University Course Timetabling Problem is an active and complex field with broad implications at most educational institutions. The current gap between literature and real-world applications is among the foremost challenges in this domain. This gap can be attributed to the solvers’ lack of generality and/or inconsistency in problem definition. In an attempt to tackle this issue, the Student Sectioning Problem has been incorporated into the latest International Timetabling Competition benchmark dataset. This paper reviews, categorizes, and summarizes the approaches addressing the University Course Timetabling with Student Sectioning Problem. Furthermore, the strengths and limitations of these methods are analyzed. Challenges, gaps, and possible future work are also presented.

Diabetes is a chronic disease characterized by a decrease in pancreatic insulin production. The immune system will be harmed due to this condition, which will raise blood sugar levels. However, early detection of diabetes enables patients to begin treatment on time, therefore reducing or eliminating the risk of severe consequences. One of the most significant challenges in the healthcare unit is disease diagnosis. Traditional techniques of disease diagnosis are manual and prone to inaccuracy. This paper proposed an approach for diagnosing diabetes using the adaptive neuro-fuzzy inference system (ANFIS) based on Pima Indians diabetes dataset (PIDD). The three stages of the proposed approach are pre-processing classification and evaluation. Normalization, imputation, and anomaly detection are part of the pre-processing stage. The pre-processing was done by normalizing the data, replacing the missing values, and using the local outlier factor (LOF) technique. In the classification stage, ANFIS classifiers were trained using the hybrid learning algorithm of the neural network. Finally, the evaluation procedures use the last stage’s sensitivity, specificity, and accuracy metrics. The obtained classification accuracy was 92.77%, and it seemed rather promising compared to the other classification applications for this topic found in the literature.

Harmony search algorithm (HSA) is relatively considered as one of the most recent metaheuristic algorithms. HSA is a modern - nature algorithm that simulates the musicians’ natural process of musical improvisation to enhance their instrument’s note to find a state of pleasant (harmony) according to aesthetic standards. Lots of variants of HSA have been suggested to tackle combinatorial optimization problems. They range from hybridizing some components of other metaheuristic approaches (to improve the HSA) to taking some concepts of HSA and utilizing them to improve other metaheuristic methods. This stud y reviews research pertaining to parameter settings of HSA and its applications to efficiently solve hard combinatorial optimization problems.

Analysis and fast streamflow forecasting are essential. Reliable predicting for river flow, as per the major source of usable water, which can be a crucial factor in the drought analysis and construction of waterrelated infrastructures. Data-driven and hybrid methods are increasingly being used to address the nonlinear and variable components of hydraulic processes. In this paper, a streamflow forecasting model is built utilizing Neural Network (NN) and Wavelet Transform (WT) at Western Australia for Ellen Brook River with the application of Railway Parade station. Initially, the sequences of signals are applying to the wavelet to be evaluated at several levels and extract a sequence of different features from the chosen output in the wavelet. Then, the obtained output is presented to the neural network for tuning to get the best intermittent streamflow forecasting. The existing input and structures are designed for streamflow forecasting. The proposed model has a better performance compared to the previous models. The proposed model is beneficial for application of forecasts to examine the relation between the characteristics of river flow, optimal decomposition degree, data duration, and the precise wavelet transform form.

Missing values are one of the common incidences that occurs in healthcare datasets. Its existence usually leads to undesirable results while conducting data analysis using machine learning methods. Recently, researchers have proposed several imputation approaches to deal with missing values in real-world datasets. Moreover, data imputation assists us to build a high-performance machine learning models to discover patterns in healthcare data that provides top-notch insights for a higher quality decision-making. In this paper, we propose a new imputation approach using Extremely Randomized Trees (Extra Trees) of machine learning ensemble learning methods named (ExtraImpute) to tackle numerical missing values in healthcare context. The proposed method has the ability to impute both continuous and discrete data features. This approach imputes each missing value that exists in features by predicting its value using other observed values in the dataset. To evaluate the efficiency of our algorithm, several experiments are conducted on five different benchmark healthcare datasets and compared to other commonly used imputation methods, viz. missForest, KNNImpute, Multivariate Imputation by Chained Equations (MICE), and SoftImpute. The results were validated using Root Mean Square Error (RMSE) and Coefficient of Determination (R2) scores. From these results, it was observed that our proposed algorithm outperforms existing imputation techniques.

Limited approaches have been applied to Arabic sentiment analysis for a five-point classification problem. These approaches are based on single task learning with a handcrafted feature, which does not provide robust sentence representation. Recently, hierarchical attention networks have performed outstandingly well. However, when training such models as single-task learning, these models do not exhibit superior performance and robust latent feature representation in the case of a small amount of data, specifically on the Arabic language, which is considered a low-resource language. Moreover, these models are based on single task learning and do not consider the related tasks, such as ternary and binary tasks (cross-task transfer). Centered on these shortcomings, we regard five ternary tasks as relative. We propose a multitask learning model based on hierarchical attention network (MTLHAN) to learn the best sentence representation and model generalization, with shared word encoder and attention network across both tasks, by training three-polarity and five-polarity Arabic sentiment analysis tasks alternately and jointly. Experimental results showed outstanding performance of the proposed model, with high accuracy of 83.98%, 87.68%, and 84.59 on LABR, HARD, and BRAD datasets, respectively, and a minimum macro mean absolute error of 0.632% on the Arabic tweets dataset for five-point Arabic sentiment classification problem.

In diagnosing coronary artery disease, measurement of the cross-sectional area of the lumen, maximum and minimum diameter is very important. Mainly, it will be used to confirm the diagnosing, to predict the stenosis if any and to ensure the size of the stent to be used. However, the measurement only offers by the existing software and some of the software needs human interaction to complete the process. The purpose of this paper is to present the algorithm to measure the region of interest (ROI) on intravascular ultrasound (IVUS) using an image processing technique. The methodology starts with image acquisition process followed by image segmentation. After that, border detection for each ROI was detected and the algorithm was applied to calculate the corresponding region. The result shows that the measurement is accurate and could be used not only for IVUS but applicable to solid circle and unsymmetrical circle shape.

The reliable forecasting of river flow plays a key role in reducing the risk of floods. Regarding nonlinear and variable characteristics of hydraulic processes, the use of data-driven and hybrid methods has become more noticeable. Thus, this paper proposes a novel hybrid wavelet-neural network (WNN) method with feature extraction to forecast river flow. To do this, initially, the collected data are analyzed by the wavelet method. Then, the number of inputs to the ANN is determined using feature extraction, which is based on energy, standard deviation, and maximum values of the analyzed data. The proposed method has been analyzed by different input and various structures for daily, weekly, and monthly flow forecasting at Ellen Brook river station, western Australia. Furthermore, the mean squares error (MSE), root mean square error (RMSE), and the Nash-Sutcliffe efficiency (NSE) is used to evaluate the performance of the suggested method. Furthermore, the obtained findings were compared to those of other models and methods in order to examine the performance and efficiency of the feature extraction process. It was discovered that the proposed feature extraction model outperformed their counterparts, especially when it came to long-term forecasting.

Requirement reuse came as an alternative method to facilitate the requirement engineering process. It is an approach of making use of existing requirement documents in order to reduce effort requirement elicitation and analysis during software development lifecycle. There are various research works had been done on the implementation of requirement reuse and many had sought to provide a better practice of it. However, the practice of requirement reuse is still uncertain. There is a lack of study to determine the current state of requirement reuse practice in Malaysia. Therefore, an instrument is designed and developed to determine requirement reuse process, challenges, component, and its support tools. This paper describes the design of the instrument used for the survey on requirement reuse practice and the assessment of its reliability and validity. Cronbach’s alpha test was used to check on its reliability meanwhile hetero-monotrait ratio (HTMT) criterion was used to check on the discriminant validity. The result from both of the tests shows the item in the questionnaire has a high level of consistency and established discriminant validity.Therefore, based on the results, there are no modifications or amendments made to the questionnaire. Hence, the developed instrument is reliable and has sufficient discriminant validity for further work.

Clustering, an unsupervised method of grouping sets of data, is used as a solution technique in various fields to divide and restructure data to become more significant and transform them into more useful information. Generally, clustering is difficult and complex phenomenon, where the appropriate numbers of clusters are always unknown, comes with a large number of potential solutions, and as well the datasets are unsupervised. These problems can be addressed by the Multi-Objective Particle Swarm Optimization (MOPSO) approach, which is commonly used in addressing optimization problems. However, MOPSO algorithm produces a group of non-dominated solutions which make the selection of an “appropriate” Pareto optimal or non-dominated solution more difficult. According to the literature, crowding distance is one of the most efficient algorithms that was developed based on density measures to treat the problem of selection mechanism for archive updates. In an attempt to address this problem, the clustering-based method that utilizes crowding distance (CD) technique to balance the optimality of the objectives in Pareto optimal solution search is proposed. The approach is based on the dominance concept and crowding distances mechanism to guarantee survival of the best solution. Furthermore, we used the Pareto dominance concept after calculating the value of crowding degree for each solution. The proposed method was evaluated against five clustering approaches that have succeeded in optimization that comprises of K-means Clustering, MCPSO, IMCPSO, Spectral clustering, Birch, and average-link algorithms. The results of the evaluation show that the proposed approach exemplified the state-of-the-art method with significant differences in most of the datasets tested.

Abundant of studies and intellectual efforts have been invested particularly in determining the influencing and inhibiting factors towards effective knowledge transfer. The intriguing aspects of knowledge transfer have enabled it to be studied and examined from diverse angles and multi perspectives. Many however, tend to overlook the essence of effective knowledge transfer that is, the utilization of knowledge being transferred. For knowledge-intensive environment such as software development project, knowledge integration which encompasses activities of transferring various knowledge across different software processes often takes place. Best case scenario would be a generous opportunity to learn from each other, creating new knowledge and ultimately delivering quality software. But many failed to reap its advantage resultant from its impulsive nature and lack of methods to verify use of knowledge when the transferhappens. Alarmed by this situation, we proposed PKAMI, a model for characterizing knowledge utilization thus enables the verification of effective knowledge transfer. In our previous study we focused into software architecture development. In this paper, we attempt to investigate knowledge transfer a step further, which is in the process of designing the software to be developed for a project, carried out by our students and supervised by our industrial partners. Our aim is to determine the occurrence of effective knowledge transfer by characterizing the knowledge use during software design process. We believe that our effort put forth in this research would be an extremely significant contribution to software engineering as well as knowledge management research.

A set of questionnaire was designed to study the current state of requirement reuse practices in Malaysia. This paper describes the design of the survey on requirement reuse practice instrument, and the assessment of its reliability and validity. Cronbach’s Alpha test was used to check on its reliability and respondents’ feedback, to assess the level of understanding, the level of difficulty in responding and the level of relevancy to the subject area including the duration taken to complete the questionnaire. Results from the feedback suggested three main issues that need to be looked upon as a way to improve the instrument design. Future efforts will focus on improving the structure and contents of the questionnaire in order to achieve higher reliability and better number of responses to study on requirement reuse.

The complexities and tangles of Arabic dialect in orthography and morphology typically make the sentimental analysis quite challenging. Moreover, most of the classification approaches have addressed this problem based on hand-crafted features. Since the Arabic language has multi-dialects and the language has no word-based order, the extraction process and the classification tasks are more difficult and time consuming. Deep neural network approaches applied to the Arabic language colloquial are very limited. These deep learning approaches typically comprise a structure that is very complex for small quantities of data. The structures are based on wide convolutional networks that are not capable of capturing the entire semantic and sentiment features for Arabic dialects. In this paper, a narrow structure of the convolutional neural network (CNN) has been proposed in order to obtain the tweets representations and classify the Arabic tweets into five, three and two polarities. Sensitivity analysis has been conducted to evaluate the impact of various combination structural properties, such as the number of convolutional filters, pooling size, and filter size on the classification performances. The proposed Arabic narrow convolutional neural network (NCNN) has captured the entire semantic and sentiment information contained in the tweet by maximizing the features of the detector's range. The NCNN performances were estimated to be at its optimum when structured by three convolutional layers, each one followed by the max pooling layer. The model has been developed without using lexicon resources and lexical features or augmented the dataset with extra training data. The narrow model is the first baseline model for Arabic dialects sentiment classifications for a sentence level as it is the first narrow CNN model addressing the Arabic Dialect tweets. NCNN model achieved the lowest macro average mean absolute error (MAE M ) for five polarity and higher Macro average recall (P)

Training set is the key role player that can improve the performance of any classification task. Different techniques and methods are being applied to generate training set depending on its area of application. Researchers in data science and semantic web community use different kind of training sets generated to improve the performances of classifications and information retrieval capability. Operational Training Set Generator (TSG) should always solve a minimum of two issues; (1) it must address the computational cost in producing a reasonable outcome, thereby reducing the computational cost in the whole system. The runtime of TSG is near linear as in blocking approach and (2) it must produce the qualitative training sets. We use LogTfIdf as the cosine similarity function of two given vectors to produce Bag of Words (BoW); the tokenizer is developed to specially take care of delimiters that often come across URIs and other RDF essentials. We evaluated our UTSG on nine cross-domain benchmark ontologies publically available in OAEI website. The results obtained shows that our UTSG outperforms the two baseline TSGs previously developed to address similar problem.

Dengue has become rapidly expanding and significant public health problem in tropical and subtropical regions. In severe cases, people infected with dengue may experience severe bleeding, shock and death. Thus, increasing dengue fever (DF) can be very serious, potentially life threatening and becoming global threat. Therefore, this research aimed to develop an accurate model that could better detect early signs and symptoms of dengue fever and develop a practical system for self-notification of the disease. Two techniques were applied to provide early self-notification to the patients namely the fuzzy expert system and data mining technique. The rules of dengue diagnosis are developed based on an interview with a medical doctor and those rules will be applied in an expert system using a fuzzy logic. However, before applying the extracted rules, the accuracy of rules was tested by data mining tool. This research applies the methodology to dengue related-data from a hospital and compares the rules to the training dataset by Multilayer Perceptron network. Furthermore, the finding showed that the accuracy of result for self-diagnosis of dengue symptoms produce a reliable result.

Today, most universities worldwide have initiated the green campus campaign. In line with the campaign intention to sustain and create healthy learning environment, we developed a system, which aims to facilitate the management of recycle program as part of green campus initiative. Smart Green Management System (SGMS) is a mobile crowdsourcing-based system with a capability to help fast-managing and coordinating recycling effort between the green university council and green city council. This system allows access from any mobile devices at anytime and anywhere. Centralized participatory crowdsourcing method is the key approach in developing this system. In this approach, the crowd is tasked with finding and collecting information into a common location and format. It is ideal for information gathering, organization, and reporting problems, which are demonstrated by SGMS.

Software architecture development is a creative process that requires integration of different knowledge expertise and functionalities hence, open up wide opportunities for those involved to learn from each other. Nevertheless, many assertions have been made about the flow of knowledge transfer being unclear and abstruse therefore led to failure in maximizing knowledge transfer benefits. We respond to this assertions by proposing a model for characterizing the utilization of knowledge in order to verify the occurrence of successful knowledge transfer. The model is derived based on PKAMI, a strategy we invented that comprises of 5 key steps. We further presented a simple case study to show our implementation of the model proposed. The significance of this research is that the model will be able to assist all software practitioners to verify their engagement in knowledge transfer as well as to better strategize on improving and keep on producing quality software project deliverables. We are also aiming at elevating the essence of knowledge utilization by encouraging those involved in development to find ways and opportunities to learn from others’ experiences.

Nowadays, many disciplines and technologies are focusing on the aspect of sustainability in the development of hardware and application. There are challenges to find useful sustainability components for the information system design in order to develop green Information System (IS). This paper proposed components of green system design for achieving IS sustainability. The components involved are people, hardware, data center and network in system design process, which are beneficial to the stake-holders that contribute to economic, social and environments aspects sustainability for future development.

An expert system is vital in flood warning to determine the decision making output in order for a proper control of the input consisting of river level and rainfall. The river level and rainfall input is naturally stochastic, uncertain and unpredictable. Therefore, a fuzzy logic model of the flood warning expert system has been designed for user to obtain information about flood. This fuzzy model integrates multi-agent and ontology and expected to handle both uncertainty and accuracy issues whereas previous researches focus on handling either uncertainty or accuracy issues only. Simulation of outcome based on river level and rainfall is presented in this study.

Having good cancer classifiers are crucial in order to give the most effective and cost saving treatments for patients. Microarray is one of the vital tools in cancer studies, as it allows the discovery of gene expression patterns and promises better accuracy of cancer classification. This paper presents an associative classification framework for microarray data. The proposed framework combined the strength of both filter method and association rule mining. The experimental results showed that the selected gene subsets from generated association rules can improve the accuracy and interpretability of classifiers.

Merit2U is a web-based system to manage merit points for students' college application. This system provides better merit point management where the points are calculated automatically and electronically using web-based application.

Adaptive detection is the learning ability to detect any changes in patterns in intrusion detection systems. In this paper, we propose combining two techniques in feature selection algorithm, namely consistency subset evaluation (CSE) and DDoS characteristic features (DCF) to identify and select the most important and relevant features related DDoS attacks. The proposed technique is trained and tested using the NSL-KDD 2009 dataset and compared with the traditional features selection method such as Information Gain, Gain Ratio, Chi-squared and Correlated features selection (CFS). The result shows that the combined CSE with DCF model overcomes the drawback of traditional feature selection technique such as avoid over-fitting, long training time and improved efficiency of detections. The adaptive model based on this technique can reduce computational complexity to analyze the data when attack occurs.

Early prediction of diseases especially dengue fever in the case of Malaysia, is very crucial to enable health authorities to develop response strategies and context preventive intervention programs such as awareness campaigns for the high risk population before an outbreak occurs. Some of the deficiencies in dengue epidemiology are insufficient awareness on the parameter as well as the combination among them. Most of the studies on dengue prediction use standalone models which face problem of finding the appropriate parameter since they need to apply try and error approach. The aim of this paper is to conduct experiments for determining the best network structure that has effective variable and fitting parameters in predicting the spread of the dengue outbreak. Four model structures were designed in order to attain optimum prediction performance. The best model structure was selected as predicting model to solve the time series prediction of dengue. The result showed that neighboring location of dengue cases was very effective in predicting the dengue outbreak and it is proven that the hybrid Genetic Algorithm and Neural Network (GANN) model significantly outperforms standalone models namely regression and Neural Network (NN).

Problem-Based Learning (PBL) is an approach that incorporates soft skill with content learning and independent learning style as opposed to the teacher-centered learning practiced in universities. However, its implementation is time consuming and burdensome, involving countless paper-based assessments to be graded. Therefore MyVetPBL™, an e-learning tool for veterinary students as a paperless initiative to combat these problems was developed using a Software Development Life Cycle (SDLC) and implemented using Java, PHP language and MySQL database. The integration of e-learning tool with the PBL approach will hopefully enhance teaching and learning in undergraduate veterinary education in Universiti Putra Malaysia.

In this paper, a flood prediction conceptual model that integrates software agent, ontology and fuzzy logic is proposed. The agent role is to provide information on flood based on river level and rainfall by giving notification to users. The ontology organizes flood knowledge in order to support agent communication. Fuzzy logic will be applied to predict uncertain whether situations.

Rubrics provide standard isation in evaluation by specifying guidelines to assign marks according to predetermined evaluation criteria. Rubrics-based evaluation for the final year project in computer science has typically focused on the proposal writing, seminar, prototype development, logbook maintenance and dissertation writing. However, the evaluation has often overlooked the innovativeness of the project and the entrepreneurship skills of the candidates. This article addresses the gap in implementation of the final year project and introduces a framework of innovation-enhanced rubrics-based evaluation for the final year project. The new assessment framework emphasises entrepreneurship and innovation skills, which are realised as cognitive, affective, psychomotor and soft skills, as well as life-long learning characteristics in the course implementation. The new framework has resulted in candidates who have acquired the holistic attributes of excellent fresh graduates which are important and highly regarded by employers.

Survival analysis describes the analysis of data that corresponds to the time from when an individual enters a study until the occurrence of some particular event or end-point. It is most commonly used in the context of modelling survival (or disease-free interval time) in medical contexts, often concerned with the comparison of survival for different combinations of risk factors and/or treatments. Analytical methods which are transparent to the clinicians in understanding and explaining individual inference need to be considered when dealing with such medical data. In this paper, we present a framework for modelling survival utilising the application of the ANFIS fuzzy inference system. In this framework, alternative methods of partitioning the input space can be selected to define the membership functions, for example by using expert knowledge, equalizer partitioning, fuzzy c-means clustering, or the combination of these techniques. Further, the rule base can be established by enumerating all possible combinations of membership functions of all inputs. After the initialisation of the fuzzy inference structure, the replication data (until time to event) will be subject to be trained using the gradient descent and nonnegative least square algorithm to estimate the conditional event probability. This framework is validated over a novel dataset of patients following operative surgery for ovarian cancer. We demonstrate that the proposed framework can be successfully applied to estimate the hazard and survival curves between different prognostic factors, and model survival times, while providing models with explicit explanation capabilities.

The physical and mechanical properties of bamboo have been widely studied but information of these properties at the nodes is still lacking. The presence of node in a bamboo split may affect the mechanical strength of the material. To enable usage of bamboo in a longer length, understanding the mechanical properties and behaviour of the bamboo at the node is crucial. This study used 4-year-old Gigantochloa scortechinii bamboo. The physical and mechanical properties at the nodal and internodal sections of the bamboo split were tested in green and air dry conditions. The orientations of the bamboo splits with the periphery of the bamboo split oriented facing downwards (referred to as compression) and facing upwards (referred to as tension) were taken into consideration during the mechanical test. Results suggested avoiding orientating the bamboo splits with the peripheral skin positioned at the bottom as it reduced the mechanical properties. The failure behaviour of bamboo splits at the node and internode were evaluated and compared.