Computational Intelligence in Information Systems

In a previous work, suitable initialisation techniques were incorporated with semi-supervised Fuzzy c-Means clustering (ssFCM) to improve clustering results on a trial and error basis. In this work, we present a single fully-automatic version of an existing semi-supervised Fuzzy c-means clustering framework which uses genetically-modified prototypes (ssFCMGA). Initial prototypes are generated by GA to initialise the ssFCM algorithm without experimentation of different initialisation techniques. The framework is tested on a real, biomedical dataset NTBC and on the Arrhythmia UCI dataset, using varying amounts of labelled data from 10 % to 60 % of the total data patterns. Different ssFCM threshold values and fitness functions for ssFCMGA are also investigated (sGAs). We used accuracy and NMI to measure class-label agreement and internal measures WSS, BSS, CH, CWB, DB and DU to evaluate cluster quality of the clustering algorithms. Results are compared with those produced by the existing ssFCM. While ssFCMGA and sGAs produced slightly lower agreement level than ssFCM with known class labels based on accuracy and NMI, the other six measurements showed improvement in the results in terms of compactness and well-separatedness (cluster quality), particularly when labelled data are low at 10 %. Furthermore, the cluster quality are shown to further improve using ssFCMGA with a more complex fitness function (sGA2). This demonstrates the application of GA in ssFCM improves cluster quality without exploration of different initialisation techniques.

Many models have been implemented in the energy sectors, especially in the electricity load consumption ranging from the statistical to the artificial intelligence models. However, most of these models do not consider the factors of uncertainty, the randomness and the probability of the time series data into the forecasting model. These factors give impact to the estimated model’s coefficients and also the forecasting accuracy. In this paper, the fuzzy random auto-regression model is suggested to solve three conditions above. The best confidence interval estimation and the forecasting accuracy are improved through adjusting of the left-right spreads of triangular fuzzy numbers. The yearly electricity load consumption of North-Taiwan from 1981 to 2000 are examined in evaluating the performance of three different left-right spreads of fuzzy random auto-regression models and some existing models, respectively. The result indicates that the smaller left-right spread of triangular fuzzy number provides the better forecast values if compared with based line models.

In recent years a large number of evolutionary and other population based heuristics were proposed in the literature for solving NP-hard optimization problems. In 2015 we presented a Discrete Bacterial Memetic Evolutionary Algorithm (DBMEA) for The Traveling Salesman Problem. It provided results tested on series of TSP problems. In this paper we present an improved version of the DBMEA algorithm, where the local search is accelerated, which is the most time consuming part of the original DBMEA algorithm. This modification led to a significant improvement, the runtime of the improved DBMEA was 5–20 times shorter than the original DBMEA algorithm. Our DBMEA algorithms calculate real value costs better than integer ones, so we modified the Concorde algorithm be comparable with our results. The improved DBMEA was tested on several TSPLIB benchmark problems and other VLSI benchmark problems and the following values were compared: - optima found by the improved DBMEA heuristic and by the modified Concorde algorithm with real cost values - runtimes of original DBMEA, improved DBMEA and modified Concorde algorithm. Based on the test results we suggest the use of the improved DBMEA heuristic for the more efficient solution of TSP problems.

The determination of stampede occurrence through abnormal behaviors is an important research in context-awareness using individual activity recognition (IAR). An application such as an intelligent smartphone for crowd monitoring using inbuilt sensors is used. Meanwhile, there are few algorithms to recognize abnormal behaviors that can lead to a stampede for mitigation of crowd disasters. This study proposed an improved stampede prediction model which can facilitate abnormal detection with k-means. It can identify cluster areas among a group of people to know susceptible places that can help to predict stampede occurrence using IAR with the help of geographical positioning system (GPS) and accelerometer sensor data. To achieve this, two research questions were formulated and answered in this paper. (i) How to determine crowd of people in an area? (ii) How to know when stampede will occur in the identified area? The experimental results on the proposed model with decision tree (DT) algorithm shows an improved performance of 98.6 %, 97.7 % and 10.9 % over 94.4 %, 95 % and 18 % in the baselines for specificity, accuracy and false-negative rate (FNR) respectively thereby reducing high false negative alarm.

This paper investigates the accuracy of five state-of-the-art machine learning techniques — decision tree J48, nearest neighbors, k-nearest neighbors (k-NN), backpropagation neural network, and naive Bayes — for image-based snake species identification problem. Conventionally, snake species identification is conducted manually based on the observation of the characteristics such head shape, body pattern, body color, and eyes shape. Images of 22 species of snakes that can be found in Malaysia were collected into a database, namely the Snakes of Perlis Corpus. Then, an intelligent approach is proposed to automatically identify a snake species based on an image which is useful for content retrieval purpose where a snake species can be predicted whenever a snake image is given as input. Our experiment shows that backpropagation neural network and nearest neighbour are highly accurate with greater than 87 % accuracy on CEDD descriptor in this problem.

Bus transit is not popular in Brunei partly due to high ownership of private cars and this will lead to severe traffic congestion in the future. This paper discusses the design of a mobile application to sustain an incentive scheme for public bus transportation in Brunei. It supports a case study on whether an incentive scheme has impact to increase bus transit ridership under the prevailing transport conditions. The application has a front-end mobile app to collect dates and time stamps of each ride and a back-end infrastructure to manage stakeholder details, incentive schemes, bus routes and a repository for travel time stamps.

Near field communication (NFC) is a form of short range contactless and wireless communication between devices, a subset of RFID, with a much shorter communication range for security purposes. Any objects can become a passive device by tagging a NFC tag on them. In the past, RFID and smart card was a feasible choice for bus fare/token ticketing as their tag can cost as low as 10 cents/piece, being also the average price for a NFC tag. However, their tag readers are expensive, such as passive RFID and smart card reader, and not universal. Thus, this project proposes an economical and portable ticketing system, named mBus, designed to (1) simplify payments for bus rides and (2) low-cost fare/token tracking using NFC. The proposed mBus consist of (1) bTag, (2) bReader, and (3) bData. The bTag are able to (1) store information, and (2) communicate with an active reader device, bReader. User can purchase a bTag, being a NFC tag, and a bus pass. bTag contain crucial information such as user information and remaining token/fare. A mobile application, named as bReader, allows the bus driver to (1) read bTag, (2) act as a payment gateway, and (3) create new users. The system record user data in bData, being a remote SQL database using PHP web pages. The system is being designed as mobile application since mobile devices has huge market in the portable devices industry and recent mobile system has NFC chips installed in them, i.e. anyone who desires to use this bus fare/token ticketing system can find and obtain inexpensive NFC-embedded mobile device easily.

Trust between human and robot is one of the crucial issues in robot-based therapy. It is highly important to provide a clearer and richer understanding and also to answer the questions why trust occurs in machines and how it can maintain successful interaction. In this paper, an agent based model for trust dynamics in short-term human-robot interaction is discussed and formally analysed. Three different cases were implemented to simulate various scenarios that explain the development of trust during short-term human-robot interaction; namely, (1) high level of trust, (2) moderate level of trust, and (3) low level of trust. Furthermore, simulation traces for fictional characters under different cases have pointed out realistic behaviours as existed in the literature. The developed model was verified by using mathematical (stability analysis) and automated verification (Temporal Trace Language).

This paper presents the development of an ambient agent model (software agent) as an initial step to develop a reading companion robot to sup-port reading performance. This ambient agent model provides detailed knowledge (human functioning) about reader’s dynamics states. Based on this human functioning knowledge, a robot will be able to reason about reader’s conditions and provides an appropriate support. Several simulation traces have been generated to illustrate the functioning of the proposed model. Furthermore, the model was verified using an automated trace analysis and the results have shown that the ambient agent model satisfies a number of related properties as presented in related literatures.

The aim of this paper is to investigate the factors influencing student’s acceptance and attitude towards using 3D virtual learning spaces for education. Extended Technology Acceptance Model has been utilized as its hypothetical premise by incorporating self-efficacy and perceived enjoyment as new external variables. The model is tested through a survey administered to 85 students who took an interest in using the 3D virtual learning spaces. We conducted a regression analysis to examine the potential influence of independent variables on the acceptance and attitude towards using 3D virtual learning spaces. Our result showed that attitude towards using was the significant influence on behavior intention to use. In addition, the reconciliation of self-efficacy and perceived enjoyment are also significant antecedents to perceived ease of use and perceived usefulness. This study confirms that self-efficacy, perceived enjoyment, perceived ease of use, and perceived usefulness are important variables of acceptance and attitude towards using 3D virtual learning spaces.

One of the significant problems in classification is class noise which has numerous potential consequences such as reducing the overall accuracy and increasing the complexity of the induced model. Subsequently, finding and eliminating misclassified instances are known as important phases in machine learning and data mining. The predictions of classifiers can be applied to detect noisy instances, inconsistent data and errors, what is called classification filtering. It creates a new set of dataset to develop a reliable and precise classification model. In this paper we analyze the effect of class noise on six supervised learning algorithms. To evaluate the performance of the classification filtering algorithms, several experiments were conducted on six real datasets. Finally, the noisy instances are removed and relabeled and the performance was then measured using evaluation criteria. The findings of this study show that classification filtering have a potential capability to detect class noise.

\(R^2\), despite being a widely used goodness-of-fit measure for linear regression shows erratic behavior in presence of data contamination. Several alternate measures have been proposed that show some improvement under specific conditions. However, no single universal measure exists as such that can be used to assess and compare performance of linear regression models without being concerned about composition of data. This paper proposes a new robust \(R^2\) measure that is found to work better than existing measures across scenarios. Performance superiority has been demonstrated using extensive simulation results and three real publicly available datasets. Proposed methodology also shows significant improvement in outlier detection and comparable performance to other established methods for robust linear regression.

Using StockProF developed in our previous work, we are able to identify outliers from a pool of stocks and form clusters with the remaining stocks based on their financial performance. The financial performance is measured using financial ratios obtained directly or derived from financial reports. The resulted clusters are then profiled manually using mean and 5-number summary calculated from the financial ratios. However, this is time consuming and a disadvantage to novice investors who are lacking of skills in interpreting financial ratios. In this study, we utilized class association rule mining to overcome the problems. Class association rule mining was used to form rules by finding financial ratios that were strongly associated with a particular cluster. The resulted rules were more intuitive to investors as compared with our previous work. Thus, the profiling process became easier. The evaluation results also showed that profiling stocks using class association rules helps investors in making better investment decisions.

Many clinical data suffer from data imbalance in which we have large number of instances of one class and small number of instances of the other. This problem affects most machine learning algorithms especially decision trees. In this study, we investigated different undersampling and oversampling algorithms applied to multiple imbalanced clinical datasets. We evaluated the performance of decision tree classifiers built for each combination of dataset and sampling method. We reported our experiment results and found that the considered oversampling methods generally outperform undersampling ones using AUC performance measure.

Behind Internet of Things (IoT) system, there are constrained devices and protocols that handle all the communication in the system. Constrained devices are equipped with sensor and communication capabilities to allow them to send data over the network. There are limitations on these constrained devices as they have limited resources such as processing power, memory and power consumption. In order to fit the needs for IoT systems, different types of protocols have been developed. These protocols lie in a communication protocol stacks that are from the Application layer, Transport layer, Network Layer and Network access layer. These protocols are designed to cater for the need of the systems to run smoothly with the limited resources available for the constrained devices. For this paper, the focus lies on two (IoT) protocols on the application layer: Hypertext Transfer Protocol (HTTP) and Constrained Application Protocol (CoAP). It extends how the protocol structures the message format, communication establishment and how request is handled from the client. The study is designed on different test beds based on performance factor to meet the requirement of the device’s resources. The results and analysis of this study contributes to the findings of the performance where CoAP is faster than HTTP with smaller data. The study strengthens the use of CoAP for constrained devices in relation to the limited resources mentioned before thus contributing to how data are managed in any IoT environment.

Considering ever increasing importance of traffic sharing am- ong the users connected with different Optical Network Units (ONUs) of a TDM-Passive Optical Network (TDM-PON), to date, different TDM-PON architectures have been introduced. These architectures facilitate traffic flow among the ONUs directly along with traffic flow between an ONU and the Optical Line Terminal (OLT), which is the centralized intelligence of a TDM-PON system. The TDM-PON that facilitates direct communication among different ONUs under a single TDM-PON system is namely called TDM-PON Internetworking architecture. In this paper, we come up with a novel operational mechanism for one of the well-known TDM-PON Internetworking architectures in order to improve energy saving performance. An ONU in the TDM-PON Internetworking architecture based on which we propose a novel Internetworking operational mechanism can have two modes: LAN-PON (sharing traffic with other ONUs directly without any assist from the OLT) and broadband access (facilitates OLT and ONU communication). In order to facilitate operating under these two modes, an ONU uses two low-cost Optical Switches (OSWs). We refer to this Internetworking architecture as OSW based solution. Our novel energy efficient operational mechanism allows an ONU in OSW based solution to use Energy Saving Mode (ESM). The simulation results state that the operational mechanism introduced in this paper can contribute in increasing energy saving performance of ONUs noticeably.

Black Hole attack is one of the many attacks that can occur on a MANET. It works on the network layer by dropping all incoming packets instead of forwarding them to the destinations. MANET conveys communication in a multi-hop manner from the source node to the destination. But without any security means, the effects of such attack on MANET can disrupt the performance and operations of the network. This paper describes the effect of the attack on MANET that is using AODV, OLSR and TORA routing protocols with the introduction of IPSec protocol under the influence of Black Hole attack. The simulation of the attack is achieved using Riverbed Modeler Academic Edition. Based on the Black Hole attack simulations, MANET suffered the lowest throughput when it is using TORA and the highest using OLSR routing protocol.

Learning management systems (LMS) such as LattitudeLearning, BIStrainer, Blackboard, Google Classroom and Moodle are commonly adapted by many education institutions. Most of the existing LMS are focusing on assisting teaching and learning related matters and does not take consideration in regards to the differences that existed between each individual learners. The main problem is that learners have different motivation, cognitive traits, and learning styles. This paper is examining the effect of learning styles by integrating adaptive courses into the LMS to suit according to the learner’s learning styles. The results revealed that learners exhibited different preferences in LMS environment based on different learning style. This paper focuses on taking account of the learner learning styles by incorporating adaptivity into the learning model. Based on this approach, the current Moodle based LMS has been implemented. By extending LMS with adaptivity, it will enable a support to teachers and learners. The research results are important to ensure that the courses include the features which fit to different learning styles. This will further identify the needs and characteristics of learners by responding to the learners and present them with the enhanced adaptive LMS based on the learners’ needs.

The Malaysian English Language Teaching (ELT) of English as a Second Language (ESL) has been a long debated topic from implementation of methodologies, strategies and models to the tools used to deliver English education to Malaysian learners. This paper highlights the two emerging tools namely Mobile Learning (M-learning) and Mobile Augmented Reality Learning (MAR-learning) in ELT. The aim of this study is to conduct a comparative study verifying if MAR-learning is significantly more motivating and satisfying than M-learning in ELT. This paper will first present the current trends in both technologies, followed by the development of the first MAR-learning prototype before being evaluated comparatively with the current existing M-learning application. The last section of this paper will highlight the results of the study in compliance with the generated hypotheses.

The present study is undertaken among Information and Communication Technology Centers (ICTCs) among four higher education institutions (HEIs) in Brunei Darussalam to evaluate and highlight their performance in achieving IT Governance using a performance measuring COBIT framework. The COBIT also assesses the maturity level indicators from the range of 0 to 5. The results of this preliminary study show that not all the items measuring five domains of COBIT is applicable to these ICTCs and weighted average maturity level of all these ICTs range from 1.40 to 1.72. This further highlight that the five domains of COBIT measuring IT Governance are at initial stage and the top management of these HEIs should adopt a framework and provide additional resources to their ICTs in bringing IT Good Governance. Based on the study results some recommendations are made to enhance the performance.

Recently, serious games aimed at cognitive therapy have been gaining increasing importance in general health applications, creating new possibilities for various groups, including children, to access new forms of treatment. However, to develop therapeutic serious games that stimulate the cognitive abilities of children with speech and language delay (CSLD), the needs of this group and a few other issues must be considered. Therefore, this paper aims to identify the problems affecting the cognitive functions of CSLD as well as their needs, to assist in the development of a serious game for CSLD. We conducted a preliminary study through a semi-structured interview with experts in the area of therapy. Our results indicate that CSLD indeed have major difficulties that affect the development of their cognitive abilities such as memory, attention, perception, problem solving, decision making, language, learning, and reasoning. In addition, CSLD also lack preverbal and motor skills. These findings reinforce the need to propose a model for therapeutic serious game design that stimulates the cognitive abilities of CSLD.

Advances in facial animation technology have been extended into various creative applications such as computer games, 3D animations and interactive multimedia. In this work, we explore performance-driven facial animations using a rigged bone approach. A 3D face model is rigged with bones positioned at the desired control points. The performance information for controlling these control points are extracted using the marker-based technique where color dots are painted on a performer’s face at desired positions. Facial expression performances are recorded using a low cost camera. The information from the recorded performances is extracted and mapped onto a targeted face. Hence facial expression performances of the actor can be applied to different 3D model faces, provided that they share the same control points. This approach affords reusability of the facial performances without expensive equipment. We present the creative process of our approach, discuss the outcome and the prospect of future works.

We look at generalized Delaunay graphs in the constrained setting by introducing line segments which the edges of the graph are not allowed to cross. Given an arbitrary convex shape C, a constrained Delaunay graph is constructed by adding an edge between two vertices p and q if and only if there exists a homothet of C with p and q on its boundary that does not contain any other vertices visible to p and q. We show that, regardless of the convex shape C used to construct the constrained Delaunay graph, there exists a constant t (that depends on C) such that it is a plane t-spanner of the visibility graph.

Several researchers have studied subgraph enumeration algorithms that use a compressed expression for a family of sets, called a zero-suppressed binary decision diagram (ZDD), to solve subgraph optimization problems. We have two representative approaches to manipulate ZDDs effectively. One is fundamental mathematical operations on families of sets over ZDDs. The other is a direct construction method of a ZDD that represents desired subgraphs of a graph and is called frontier-based search. In this research, we augment the approaches by proposing two new operations, called disjoint join and joint join, on family algebra over ZDDs and extending the frontier-based search to enumerate subgraphs that have a given number of vertices of specified degrees. Employing the new approaches, we present enumeration algorithms for alphabet letter graphs on a given graph. Moreover, we solve a variant of the longest path problem, called the Longest Oneway-ticket Problem (LOP), that requires computing the longest trip on the railway network of the Japan Railways Group using a oneway ticket. Numerical experiments show that our algorithm solves the LOP and is faster than the existing integer programming approach for some instances.