1. ELICITING EXPERT KNOWLEDGE IN QUANTIFYING THE HUMAN BEHAVIOUR DURING AN EMERGENCY USING EBBN PROCEDURE
Nurulhuda Ramli, Noraida Abdul Ghani, Intan Hashimah Mohd Hashim and Zulkarnain A. Hatta Universiti Sains Malaysia, 11800 Penang, Malaysia
firstname.lastname@example.org [Download PDF]
ABSTRACT: Elicitation of expert knowledge is a structured approach to consult experts on uncertain subject and where there is insufficient knowledge. Expert elicitation is most often used in identifying, developing and quantifying the unknown parameters in a causal model. In modelling human behavior during an emergency one has to deal with uncertainties with often limited or incomplete knowledge database. This paper is an extension of an earlier work that identified factors affecting human behavior during an emergency. The factors, stressful conditions, individual’s ability in assessing a danger and information regarding the threats were captured in a graphical representation called the Bayesian Network (BN). This study focuses on the quantification phase of the model by conducting an expert elicitation exercise which aims to extract the expert’s knowledge on the inter-dependencies of the factors involved. The experience builds on the semi-structured interviews with the expert who participated in the analysis to give their beliefs by quantifying the relationship of variables using a probability scale. In order to cope with eliciting a large number of probability values from the experts, an elicitation using the Bayesian Belief Network (EBBN) procedure has been carried out. The EBBN requires only a limited amount of elicited probabilities from the experts and uses piecewise linear interpolation to determine the conditional probability of the target variables. The generated probabilities obtained are then used to make inference on the model by inserting and propagating the appropriate evidences throughout the network. Result of the analysis shows that an individual would make a decision to evacuate from a dangerous situation when there is a medium level of stressful conditions experienced, which is dependent on having enough information about the threats received and a high ability in assessing the danger. The finding suggests that formal expert elicitation can support human behavior research when there is limited available knowledge. This research generated many useful insights from the experts involved in the elicitation exercise. The feedback and recommendations for enhancing future procedures with multiple experts are highlighted based on the lessons learnt. Future work of the study should test the validity and sensitivity of the network.
Keywords: Bayesian Network, EBBN, Evacuation, Expert Judgement, Human Behaviour.
2. AGENT BASED TWO BUFFER HIERARCHICAL SCHEDULING ALGORITHM FOR MULTICORE ARCHITECTURE
G.Muneeswari and E.M.Malathy
Associate Professor, SSN College of Engineering
email@example.com, firstname.lastname@example.org [Download PDF]
ABSTRACT: In the current era, we have moved from multiprocessor system to multicore system. The main travel towards multicore system is the tremendous performance achievement over processor execution. Although there are many hardware challenges imposed on this architecture, the software design also brought into the attention of designing efficient operating system, building intelligent compiler etc., Though many processor scheduling algorithms are developed, keeping all the cores in the active state is a major challenge which conventional algorithms do not implement. In this paper we propose a new agent based two buffer hierarchical scheduling algorithm that enhances the performance of the processor by 20% compared with the traditional algorithms. There are two levels in the overall design wherein in the first level all the tasks are assigned with equal priority and a buffering method is implemented. Whereas in the second level, we consider the real time task and affinity based scheduling is incorporated. For the evaluation results modified linux 2.6.21 kernel along with the FLAME tool is used. Ultimately, the overall results proves that this agent based algorithm outperforms in cpu performance and reduces average waiting time of the process by 4.5% compared with the conventional scheduling algorithms.
Keywords: agent, buffer, hierarchical scheduling, multicore, affinity
3. THREE METHODS OF ARTIFICIAL EVOLUTION FOR AUTOMATED & SEMI-AUTOMATED SYNTHESIS OF FREE-FORM 3D PRINTABLE AESTHETIC OBJECTS
Jason Teo, Ong Jia Hui, Halimah Manja and Lee Chin Kuan
Faculty of Computing & Informatics
Kota Kinabalu Campus
Universiti Malaysia Sabah
Jalan UMS, 88400 Kota Kinabalu, Sabah. [Download PDF]
ABSTRACT: Designing a 3D object is a very laborious process that usually involves significant expertise and time investment through the use of various 3D computer-aided design software. Numerous researchers have proposed mathematical formulas to automatically design aesthetic shapes in 2D space and this has led to recent efforts being done on studies which use mathematical formulas to create objects in 3D space. In this paper, we report on the use of the Gielis Superformula to automatically generate 3D object shapes through an artificial evolutionary optimization process. Free-form 3D shapes are synthesized through three different methods: (1) a fully automated single-objective approach, (2) a fully automated multi-objective approach, and (3) an interactive, semi-automated multi-objective approach. Various novel fitness functions were designed to evaluate the shapes generated by the Superformula in order to discriminate between aesthetic versus non-aesthetic shapes. Post-evolution shapes were then fabricated using 3D printing for human evaluation. The results demonstrate that the proposed approaches are indeed feasible in terms automating part of or even the entire design process for synthesizing free-form 3D printable that are aesthetically pleasing.
Keywords: Evolutionary 3D Art, Evolutionary Optimization, Automatic 3D Shape Generation, Gielis Superformula, Computational Aesthetics.
4. IMPROVED FUZZY-PI CONTROL SCHEME FOR POWER FLOW OF DISTRIBUTED GENERATION
Azuki Abdul Salam, Nik Azran Ab Hadi, Fatimah Zaharah Hamidon and Ismail Adam
Universiti Kuala Lumpur-British Malaysian Institute
Universiti Teknikal Malaysia Melaka,
Universiti Kuala Lumpur-British Malaysian Institute
Universiti Kuala Lumpur-British Malaysian Institute
email@example.com [Download PDF]
ABSTRACT: This paper presents the mathematical model of the Proton Exchange Membrane Fuel Cell (PEMFC) and analyzes the structure of a grid connected PEMFC generation system. In order to get better waveforms of grid current, a Fuzzy-PI controller is introduced into the grid connected PEMFC generation system. The current control scheme for grid connected PEMFC generation system is a PI controller scheme, which would lead to large transient response due to the load increases. Thus, a Fuzzy-PI control scheme is proposed in order to improve the power flow control. The PI controller parameters automatically, according to changes of system parameters. When the proposed grid connected PEMFC generation system using the Fuzzy-PI controller is simulated in Matlab/Simulink, the results show that the proposed control scheme works effectively for the power flow grid.
Keywords: PEMFC, Fuzzy-PI, Matlab/Simulink
5. DISCRETE WAVELET PACKET TRANSFORM FOR ELECTROENCEPHALOGRAM-BASED EMOTION RECOGNITION IN THE VALENCE-AROUSAL SPACE
Farzana Kabir Ahmad*and Oyenuga Wasiu Olakunle
Computational Intelligence Research Cluster,
School of Computing, College of Arts and Sciences
Universiti Utara Malaysia, 06010 UUM Sintok, Kedah, Malaysia
firstname.lastname@example.org [Download PDF]
ABSTRACT: Human emotion recognition is the key step toward innovative human-computer interactions. The advanced in computational algorithms and techniques has recently offered the promising results in recognizing human emotion. Recently, Electroencephalogram (EEG) has been shown as an effective way in identifying human emotion since it records the brain activity of human and can hardly be deceived by voluntary control. However, due to the non-linearity, non-stationary, and chaotic nature of the EEG signals, it is difficult to be examined and has been an extensive research area in the present years. Moreover, the high dimensional of the feature vectors has make the analysis task more challenging. In this research, two emotion recognition experiments were performed in order to classify human emotional states into high/low valence or high/low arousal. The first experiment was aimed to evaluate the performance of Discrete Wavelet Packet Transform (DWPT) in extracting relevant features, while the second experiment was conducted to identify the combination of electrode channels that optimally recognize emotions based on the valence-arousal model. Additionally, in this study, a leave-one-out cross validation was performed using Radial Basis Function-Support Vector Machines (RBF-SVM) as the classifier on a publically available dataset. The experimental results have shown that an average accuracy of 68.83% with average F1-score of 0.666 for valence and average accuracy of 68.83% with F1-score of 0.633 for arousal were achieved for 32 subjects. Furthermore, four frontal channels which include Fp1, Fp2, F3, and, F4 were identified significant in providing relevant information compare to the remaining 6 channels namely T7, T8, P3, P4, O1, and O2 for EEG-based emotion recognition in the valence-arousal space.
Keywords: Discrete Wavelet Packet Transform; Electroencephalogram; emotion recognition; valence-arousal model.
6. AN EFFICIENT METHOD TO PREDICT DENGUE OUTBREAKS IN KUALA LUMPUR
Duc Nghia Pham, Tarique Aziz1, Ali Kohan, Syahrul Nellis, Juraina binti Abd. Jamil,
Jing Jing Khoo, Dickson Lukose, Sazaly bin Abu Bakar and Abdul Sattar
MIMOS Berhad, Malaysia
TIDREC, University of Malaya, Malaysia
IIIS, Griffith University, Australia [Download PDF]
ABSTRACT: In recent years, there has been a surge in dengue outbreaks in Malaysia. A dengue outbreak can cause severe damages to the society. Hence, it is critical to be able to predict a dengue outbreak in advance to minimize the damage and loss. In this paper, we propose a new machine learning approach to predict the number of dengue cases in Kuala Lumpur, in particular the areas surrounding the University of Malaya (UM) Medical Centre. We identified several different factors that can contribute to a surge in the number of dengue cases that occurred near the UM Medical Centre. Apart from the daily mean temperature and daily rainfall factors that have been frequently used in other studies, we also considered the enhanced vegetation index (EVI) as an input factor to our prediction engine. We trained our linear regression model on these three factors against the number of dengue cases from 2001 to 2010. We then tested our model on the 2011 data. The experimental results showed that our approach was able to predict the number of dengue cases 16 days in advance with high accuracy.
Keywords: Dengue Outbreak Prediction, Linear Regression Model.