International Journal of Advances in Intelligent Informatics
Not a member yet
235 research outputs found
Sort by
Japanese sign language classification based on gathered images and neural networks
Full text linkThis paper proposes a method to classify words in Japanese Sign Language (JSL). This approach employs a combined gathered image generation technique and a neural network with convolutional and pooling layers (CNNs). The gathered image generation generates images based on mean images. Herein, the maximum difference value is between blocks of mean and JSL motions images. The gathered images comprise blocks that having the calculated maximum difference value. CNNs extract the features of the gathered images, while a support vector machine for multi-class classification, and a multilayer perceptron are employed to classify 20 JSL words. The experimental results had 94.1% for the mean recognition accuracy of the proposed method. These results suggest that the proposed method can obtain information to classify the sample words
Improving learning vector quantization using data reduction
Full text linkLearning Vector Quantization (LVQ) is a supervised learning algorithm commonly used for statistical classification and pattern recognition. The competitive layer in LVQ studies the input vectors and classifies them into the correct classes. The amount of data involved in the learning process can be reduced by using data reduction methods. In this paper, we propose a data reduction method that uses geometrical proximity of the data. The basic idea is to drop sets of data that have many similarities and keep one representation for each set. By certain adjustments, the data reduction methods can decrease the amount of data involved in the learning process while still maintain the existing accuracy. The amount of data involved in the learning process can be reduced down to 33.22% for the abalone dataset and 55.02% for the bank marketing dataset, respectively
Anomaly detection on flight route using similarity and grouping approach based-on automatic dependent surveillance-broadcast
Full text linkFlight anomaly detection is used to determine the abnormal state data on the flight route. This study focused on two groups: general aviation habits (C1)and anomalies (C2). Groups C1 and C2 are obtained through similarity test with references. The methods used are: 1) normalizing the training data form, 2) forming the training segment 3) calculating the log-likelihood value and determining the maximum log-likelihood (C1) and minimum log-likelihood (C2) values, 4) determining the percentage of data based on criteria C1 and C2 by grouping SVM, KNN, and K-means and 5) Testing with log-likelihood ratio. The results achieved in each segment are Log-likelihood value in C1Latitude is -15.97 and C1Longitude is -16.97. On the other hand, Log-likelihood value in C2Latitude is -19.3 (maximum) and -20.3 (minimum), and log-likelihood value in C2Longitude is -21.2 (maximum) and -24.8 (minimum). The largest percentage value in C1 is 96%, while the largest in C2 is 10%. Thus, the highest potential anomaly data is 10%, and the smallest is 3%. Also, there are performance tests based on F-measure to get accuracy and precision
Modified lambert beer for bilirubin concentration and blood oxygen saturation prediction
Full text linkNoninvasive measurement of health parameters such as blood oxygen saturation and bilirubin concentration predicted via an appropriate light reflectance model based on the measured optical signals is of eminent interest in biomedical research. This is to replace the use of conventional invasive blood sampling approach. This study aims to investigate the feasibility of using Modified Lambert Beer model (MLB) in the prediction of one’s bilirubin concentration and blood oxygen saturation value, SO2. This quantification technique is based on a priori knowledge of extinction coefficients of bilirubin and hemoglobin derivatives in the wavelength range of 440 – 500 nm. The validity of the prediction was evaluated using light reflectance data from TracePro raytracing software for a single-layered skin model with varying bilirubin concentration. The results revealed some promising trends in the estimated bilirubin concentration with mean ± standard deviation (SD) error of 0.255 ± 0.025 g/l. Meanwhile, a remarkable low mean ± SD error of 9.11 ± 2.48 % was found for the predicted SO2 value. It was concluded that these errors are likely due to the insufficiency of the MLB at describing changes in the light attenuation with the underlying light absorption processes. In addition, this study also suggested the use of a linear regression model deduced from this work for an improved prediction of the required health parameter values
Multilevel thresholding hyperspectral image segmentation based on independent component analysis and swarm optimization methods
Full text linkHigh dimensional problems are often encountered in studies related to hyperspectral data. One of the challenges that arise is how to find representations that are accurate so that important structures can be clearly easily. This study aims to process segmentation of hyperspectral image by using swarm optimization techniques. This experiments use Aviris Indian Pines hyperspectral image dataset that consist of 103 bands. The method used for segmentation image is particle swarm optimization (PSO), Darwinian particle swarm optimization (DPSO) and fractional order Darwinian particle swarm optimization (FODPSO). Before process segmentation image, the dimension of the hyperspectral image data set are first reduced by using independent component analysis (ICA) technique to get first independent component. The experimental show that FODPSO method is better than PSO and DPSO, in terms of the average CPU processing time and best fitness value. The PSNR and SSIM values when using FODPSO are better than the other two swarm optimization method. It can be concluded that FODPSO method is better in order to obtain better segmentation results compared to the previous method
A coarse-grained parallelization of genetic algorithms
Full text linkGenetic algorithms are frequently used to solve optimization problems. However, the problems become increasingly complex and time consuming. One solution to speed up the genetic algorithm processing is to use parallelization. The proposed parallelization method is coarse-grained and employs two levels of parallelization: message passing with MPI and Single Instruction Multiple Threads with GPU. Experimental results show that the accuracy of the proposed approach is similar to the sequential genetic algorithm. Parallelization with coarse-grained method, however, can improve the processing and convergence speed of genetic algorithms
Green turtle and fish identification based on acoustic target strength
Full text linkFisherman accidentally caught sea turtles in their fishnet. It could be dangerous for its population. This study measures the turtle target strength (TS) using modified echosounder. The result could be used to improve the efficiency of turtle repellent device. The experiment conducted in a hatchery fiber tank contained saline water. The Green were 1, 3, 12 and 18 years old. This study used three species of fish, which serves to distinguish the value between fish and sea turtles. TS of the animals were calculated incorporating reference targets (sphere). The echo power of the turtle was compared with the solid steel sphere which is confirmed good agreements with the theoretical values. The echo power reference by applying Fast Fourier Transform (FFT) analysis has been used in calculating TS of the animal. The time domain of the echo evaluation in different angles shows the difference in the structure of the echo signal between the tortoise's body parts. This study reveals that high echo strength is acquired from the carapace and the plastron parts. The finding also showed that there are significant differences between 3, 12, 18 years old turtles and fish in every angle measurement
Mathematics and statistics related studies in Indonesia using co-authorship network analysis
Full text linkIndonesian scholars have published a numbers of articles in numerous international publications, however, it still lags behind other Singapore, Malaysia, and Vietnam. This article performs a bibliometrics analysis and examine the collaboration network in Mathematics and Statistics related subject of scholars with Indonesian affiliation as recorded in Web of Science. In total, based on article publications during 2009-2017, 426 articles were retrieved. Bandung Institute of Technology (ITB) was the affiliation with the highest number of articles (48%) and number of authors (27%). Using Social Network Analysis to examine co-authorship networks, this research shows that the co-author network has the highest centrality in the ITB affiliation. Meanwhile, dependency of foreign affiliation is still high, shown as a high percentage (84% of all articles) of international co-authorship. Co-authorship network of Mathematics and Statistics related studies in Indonesia possesses as a scale-free network and followed the power law distribution. This research showed the achievement of Indonesian scholars of Mathematics and Statistics, and can be used to evaluate the knowledge transfer in these subjects and related areas
Biased support vector machine and weighted-smote in handling class imbalance problem
Full text linkClass imbalance occurs when instances in a class are much higher than in other classes. This machine learning major problem can affect the predicted accuracy. Support Vector Machine (SVM) is robust and precise method in handling class imbalance problem but weak in the bias data distribution, Biased Support Vector Machine (BSVM) became popular choice to solve the problem. BSVM provide better control sensitivity yet lack accuracy compared to general SVM. This study proposes the integration of BSVM and SMOTEBoost to handle class imbalance problem. Non Support Vector (NSV) sets from negative samples and Support Vector (SV) sets from positive samples will undergo a Weighted-SMOTE process. The results indicate that implementation of Biased Support Vector Machine and Weighted-SMOTE achieve better accuracy and sensitivity
Fusion noise-removal technique with modified dark-contrast algorithm for robust segmentation of acute leukemia cell images
Full text linkSegmentation is the major area of interest in the field of image processing stage. In an automatic diagnosis of acute leukemia disease, the crucial process is to achieve the accurate segmentation of acute leukemia blood image. Generally, there are three requirements of image segmentation for medical purposes, namely; accuracy, robustness and effectiveness which have received considerable critical attention. As such, we propose a new (modified) dark contrast enhancement technique to enhance and automatically segment the acute leukemic cells. Subsequently, we used a fusion 7 × 7 median filter as well as the seeded region growing area extraction (SRGAE) algorithm to minimise the salt-and-pepper noise, apart from preserving the post-segmentation edge. As per the outcomes, the accuracy, sensitivity, and specificity of this method were 91.02%, 83.68%, and 91.57% respectively