Jurnal Ilmu Komputer dan Informasi
Not a member yet
247 research outputs found
Sort by
LEAST SQUARES SUPPORT VECTOR MACHINES PARAMETER OPTIMIZATION BASED ON IMPROVED ANT COLONY ALGORITHM FOR HEPATITIS DIAGNOSIS
Get PDFMany kinds of classification method are able to diagnose a patient who suffered Hepatitis disease. One of classification methods that can be used was Least Squares Support Vector Machines (LSSVM). There are two parameters that very influence to improve the classification accuracy on LSSVM, they are kernel parameter and regularization parameter. Determining the optimal parameters must be considered to obtain a high classification accuracy on LSSVM. This paper proposed an optimization method based on Improved Ant Colony Algorithm (IACA) in determining the optimal parameters of LSSVM for diagnosing Hepatitis disease. IACA create a storage solution to keep the whole route of the ants. The solutions that have been stored were the value of the parameter LSSVM. There are three main stages in this study. Firstly, the dimension of Hepatitis dataset will be reduced by Local Fisher Discriminant Analysis (LFDA). Secondly, search the optimal parameter LSSVM with IACA optimization using the data training, And the last, classify the data testing using optimal parameters of LSSVM. Experimental results have demonstrated that the proposed method produces high accuracy value (93.7%) for the 80-20% training-testing partition
Identifying Medicinal Plant Leaves using Textures and Optimal Colour Spaces Channel
Get PDFThis paper presents an automated medicinal plant leaf identification system. The Colour Texture analysis of the leaves is done using the statistical, the Grey Tone Spatial Dependency Matrix(GTSDM) and the Local Binary Pattern(LBP) based features with 20 different colour spaces(RGB, XYZ, CMY, YIQ, YUV, , YES, , , , lms, , , HSV, HSI, IHLS, IHS, TSL, LSLM and KLT). Classification of the medicinal plant is carried out with 70\% of the dataset in training set and 30\% in the test set. The classification performance is analysed with Stochastic Gradient Descent(SGD), kNearest Neighbour(kNN), Support Vector Machines based on Radial basis function kernel(SVM-RBF), Linear Discriminant Analysis(LDA) and Quadratic Discriminant Analysis(QDA) classifiers. Results of classification on a dataset of 250 leaf images belonging to five different species of plants show the identification rate of 98.7 \%. The results certainly show better identification due to the use of YUV, and HSV colour spaces
PSNR BASED OPTIMIZATION APPLIED TO ALGEBRAIC RECONSTRUCTION TECHNIQUE FOR IMAGE RECONSTRUCTION ON A MULTI-CORE SYSTEM
Get PDFThe present work attempts to reveal a parallel Algebraic Reconstruction Technique (pART) to reduce the computational speed of reconstructing artifact-free images from projections. ART is an iterative algorithm well known to reconstruct artifact-free images with limited number of projections. In this work, a novel idea has been focused on to optimize the number of iterations mandatory based on Peak to Signal Noise Ratio (PSNR) to reconstruct an image. However, it suffers of worst computation speed. Hence, an attempt is made to reduce the computation time by running iterative algorithm on a multi-core parallel environment. The execution times are computed for both serial and parallel implementations of ART using different projection data, and, tabulated for comparison. The experimental results demonstrate that the parallel computing environment provides a source of high computational power leading to obtain reconstructed image instantaneously
Automatic Ontology Construction Using Text Corpora and Ontology Design Patterns (ODPs) in Alzheimer’s Disease
Get PDFAn ontology is defined as an explicit specification of a conceptualization, which is an important tool for modeling, sharing and reuse of domain knowledge. However, ontology construction by hand is a complex and a time consuming task. This research presents a fully automatic method to build bilingual domain ontology from text corpora and ontology design patterns (ODPs) in Alzheimer’s disease. This method combines two approaches: ontology learning from texts and matching with ODPs. It consists of six steps: (i) Term & relation extraction (ii) Matching with Alzheimer glossary (iii) Matching with ontology design patterns (iv) Score computation similarity term & relation with ODPs (v) Ontology building (vi) Ontology evaluation. The result of ontology composed of 381 terms and 184 relations with 200 new terms and 42 new relations were added. Fully automatic ontology construction has higher complexity, shorter time and reduces role of the expert knowledge to evaluate ontology than manual ontology construction. This proposed method is sufficiently flexible to be applied to other domains
FEATURE SELECTION METHODS BASED ON MUTUAL INFORMATION FOR CLASSIFYING HETEROGENEOUS FEATURES
Get PDFDatasets with heterogeneous features can affect feature selection results that are not appropriate because it is difficult to evaluate heterogeneous features concurrently. Feature transformation (FT) is another way to handle heterogeneous features subset selection. The results of transformation from non-numerical into numerical features may produce redundancy to the original numerical features. In this paper, we propose a method to select feature subset based on mutual information (MI) for classifying heterogeneous features. We use unsupervised feature transformation (UFT) methods and joint mutual information maximation (JMIM) methods. UFT methods is used to transform non-numerical features into numerical features. JMIM methods is used to select feature subset with a consideration of the class label. The transformed and the original features are combined entirely, then determine features subset by using JMIM methods, and classify them using support vector machine (SVM) algorithm. The classification accuracy are measured for any number of selected feature subset and compared between UFT-JMIM methods and Dummy-JMIM methods. The average classification accuracy for all experiments in this study that can be achieved by UFT-JMIM methods is about 84.47% and Dummy-JMIM methods is about 84.24%. This result shows that UFT-JMIM methods can minimize information loss between transformed and original features, and select feature subset to avoid redundant and irrelevant features
PERFORMANCE COMPARISON OF USART COMMUNICATION BETWEEN REAL TIME OPERATING SYSTEM (RTOS) AND NATIVE INTERRUPT
Get PDFComunication between microcontrollers is one of the crucial point in embedded sytems. On the other hand, embedded system must be able to run many parallel task simultaneously. To handle this, we need a reliabe system that can do a multitasking without decreasing every task’s performance. The most widely used methods for multitasking in embedded systems are using Interrupt Service Routine (ISR) or using Real Time Operating System (RTOS). This research compared perfomance of USART communication on system with RTOS to a system that use interrupt. Experiments run on two identical development board XMega A3BU-Xplained which used intenal sensor (light and temperature) and used servo as external component. Perfomance comparison done by counting ping time (elapsing time to transmit data and get a reply as a mark that data has been received) and compare it. This experiments divided into two scenarios: (1) system loaded with many tasks, (2) system loaded with few tasks. Result of the experiments show that communication will be faster if system only loaded with few tasks. System with RTOS has won from interrupt in case (1), but lose to interrupt in case (2)
PARTICLE SWARM OPTIMIZATION (PSO) FOR TRAINING OPTIMIZATION ON CONVOLUTIONAL NEURAL NETWORK (CNN)
Get PDFNeural network attracts plenty of researchers lately. Substantial number of renowned universities have developed neural network for various both academically and industrially applications. Neural network shows considerable performance on various purposes. Nevertheless, for complex applications, neural network’s accuracy significantly deteriorates. To tackle the aforementioned drawback, lot of researches had been undertaken on the improvement of the standard neural network. One of the most promising modifications on standard neural network for complex applications is deep learning method. In this paper, we proposed the utilization of Particle Swarm Optimization (PSO) in Convolutional Neural Networks (CNNs), which is one of the basic methods in deep learning. The use of PSO on the training process aims to optimize the results of the solution vectors on CNN in order to improve the recognition accuracy. The data used in this research is handwritten digit from MNIST. The experiments exhibited that the accuracy can be attained in 4 epoch is 95.08%. This result was better than the conventional CNN and DBN. The execution time was also almost similar to the conventional CNN. Therefore, the proposed method was a promising method.
ROBUST INTEGER HAAR WAVELET BASED WATERMARKING USING SINGULAR VALUE DECOMPOSITION
Get PDFThis paper proposed a hybrid watermarking method that used dither quantization of Singular Value Decomposition (SVD) on average coefficients of Integer Haar Wavelet Transform (IHWT). The watermark image embeds through dither quantization process on singular coefficients value. This scheme aims to obtain the higher robustness level than previous method which performs dither quantization of SVD directly on image pixels value. The experiment results show that the proposed method has proper watermarked images quality above 38dB. The proposed method has better performance than the previous method in term of robustness against several image processing attacks. In JPEG compression with Quality Factor of 50 and 70, JPEG2000 compression with Compression Ratio of 5 and 3, average filtering, and Gaussian filtering, the previous method has average Normalized Correlation (NC) values of 0.8756, 0.9759, 0.9509, 0.9905, 0.8321, and 0.9297 respectively. While, the proposed method has better average NC values of 0.9730, 0.9884, 0.9844, 0.9963, 0.9020, and 0.9590 respectively
WEB NEWS DOCUMENTS CLUSTERING IN INDONESIAN LANGUAGE USING SINGULAR VALUE DECOMPOSITION-PRINCIPAL COMPONENT ANALYSIS (SVDPCA) AND ANT ALGORITHMS
Get PDFAnt-based document clustering is a cluster method of measuring text documents similarity based on the shortest path between nodes (trial phase) and determines the optimal clusters of sequence document similarity (dividing phase). The processing time of trial phase Ant algorithms to make document vectors is very long because of high dimensional Document-Term Matrix (DTM). In this paper, we proposed a document clustering method for optimizing dimension reduction using Singular Value Decomposition-Principal Component Analysis (SVDPCA) and Ant algorithms. SVDPCA reduces size of the DTM dimensions by converting freq-term of conventional DTM to score-pc of Document-PC Matrix (DPCM). Ant algorithms creates documents clustering using the vector space model based on the dimension reduction result of DPCM. The experimental results on 506 news documents in Indonesian language demonstrated that the proposed method worked well to optimize dimension reduction up to 99.7%. We could speed up execution time efficiently of the trial phase and maintain the best F-measure achieved from experiments was 0.88 (88%)
DYNAMIC AND INCREMENTAL EXPLORATION STRATEGY IN FUSION ADAPTIVE RESONANCE THEORY FOR ONLINE REINFORCEMENT LEARNING
Get PDFOne of the fundamental challenges in reinforcement learning is to setup a proper balance between exploration and exploitation to obtain the maximum cummulative reward in the long run. Most protocols for exploration bound the overall values to a convergent level of performance. If new knowledge is inserted or the environment is suddenly changed, the issue becomes more intricate as the exploration must compromise the pre-existing knowledge. This paper presents a type of multi-channel adaptive resonance theory (ART) neural network model called fusion ART which serves as a fuzzy approximator for reinforcement learning with inherent features that can regulate the exploration strategy. This intrinsic regulation is driven by the condition of the knowledge learnt so far by the agent. The model offers a stable but incremental reinforcement learning that can involve prior rules as bootstrap knowledge for guiding the agent to select the right action. Experiments in obstacle avoidance and navigation tasks demonstrate that in the configuration of learning wherein the agent learns from scratch, the inherent exploration model in fusion ART model is comparable to the basic E-greedy policy. On the other hand, the model is demonstrated to deal with prior knowledge and strike a balance between exploration and exploitation