JUTI: Jurnal Ilmiah Teknologi Informasi
Not a member yet
407 research outputs found
Sort by
DETECTION AND CLASSIFICATION OF RED BLOOD CELLS ABNORMALITY USING FASTER R-CNN AND GRAPH CONVOLUTIONAL NETWORKS
Full text linkResearch in medical imagery field such as analysis of Red Blood Cells (RBCs) abnormalities can be used to assist laboratory’s in determining further medical actions. Convolutional Neural Networks (CNN) is a commonly used method for the classification of RBCs abnormalities in blood cells images. However, CNN requires large number of labeled training data. A classification of RBCs abnormalities in limited data is a challenge. In this research we explore a semi-supervised learning using Graph Convolutional Networks (GCN) to classify RBCs abnormalities with limited number of labeled sample images. The proposed method consists of 3 stages, i.e., extraction of Region of Interest (ROI) of RBCs from blood images using Faster R-CNN, abnormality labeling and abnormality classification using GCN. The experiment was conducted on a publicly accessible blood sample image dataset to compare classification performance of pretrained CNN models (Resnet-101 and VGG-16) and GCN models (Resnet-101 + GCN and VGG-16 + GCN). The experiment showed that the GCN model build on VGG-16 features (VGG-16 + GCN) produced the best accuracy of 95%
LOAD FORECASTING FOR DAILY LOAD OPERATIONAL PLAN USING LSTM (CASE STUDY: SOUTH SULAWESI SUB SYSTEM)
Full text linkThe electrical load required in an electricity sub-system changes every day. Electric power operators must be able to generate and distribute electricity according to consumer needs. In the Sulawesi sub-system, the power plants used are still dominated by fossil fuel generators, so that in their operations, fuel requirements need to be given serious attention. Planning a good daily electricity consumption is needed so that the fuel cost becomes optimal. In the current condition, the load forecasting for the Daily Load Operation Plan (ROH) is still based on Expert Judgment, which is different for each forecaster. With a fairly large error tolerance limit of 4%. We need a load forecasting instrument capable of better error tolerance. Forecasting methods such as ARIMA, SARIMA and ARIMAX have been used for many years. In recent years, several artificial intelligence techniques such as Neural Network and machine learning have been developed for time series analysis. And recently, more accurate forecasting results are shown by Artificial Neural Network (ANN) and Recurrent Neural Network (RNN) compared to traditional forecasting methods. Long Short Term Memory (LSTM) is a model of RNN that uses past data (Long Term) to predict current data (Short Term). Electric load in Sulawesi subsystem used as data training after normalized using min-max normalization. The LSTM model is made with different data input. Forecasting performance of each model is then evaluated based on the RMSE and MAPE values. Of the several data input models, forecasting models with daily data input show better performance than other scenarios. The MAPE and RMSE values obtained were 2.384% and 33.95, respectively
APPLICATION OF GRAPH THEORY AND WELCH-POWELL METHOD AT TRAFFIC LIGHT REGULATION
Full text linkTraffic lights are state-owned infrastructure facilities that are used to mark vehicles that must stop alternately from various directions. Traffic lights are often found at crossstreets, such as at the traffic lights on Jl. Margerejo with a long duration of red light and short green light. This study aims to obtain a traffic flow graph at the intersection of 4 Jalan Demak-Dupak Surabaya. Optimization of traffic light duration settings is very necessary on this street, because the long duration of the red light while the short duration of the green light causes the accumulation of vehicles at the intersection of Jalan Demak-Dupak Surabaya. In this study, the duration of the new traffic light was obtained, namely on Jl. Demak (North) red light 112.5 seconds and green light 37.5 seconds. For Jl. Dupak red light 84 seconds and green light 28 seconds. Jl. Demak (South) red light for 135 seconds and green light for 45 seconds. And for the axle Jl. Surabaya-Gresik red light for 84 seconds and green light for 28 seconds. The level of effectiveness of the green light is obtained by a value of 21.77% and the level of effectiveness of the red light is 6.62%
LITERATURE REVIEW IOT SOFTWARE ARCHITECTURE ON AGRICULTURE
Full text linkContext – Internet of Things (IoT) interrelates computing devices, machines, animals, or people and things that use the power of internet usage to utilize data to be much more usable. Food is one of the mandatory human needs to survive, and most of it is produced by agriculture. Using IoT in agriculture needs appropriate software architecture that plays a prominent role in optimizing the gain. Objective and Method – Implementing a solution in a specific field requires a particular condition that belongs to it. The objectives of this research study are to classify the state of the art IoT solution in the software architecture domain perspective. We have used the Evidence- Based Software Engineering (EBSE) and have 24 selected existing studies related to software architecture and IoT solutions to map to the software architecture needed on IoT solutions in agriculture. Result and Implications – The results of this study are the classification of various IoT software architecture solutions in agriculture. The highlighted field, especially in the areas of cloud, big data, integration, and artificial intelligence/machine learning. We mapped the agriculture taxonomy classification with IoT software architecture. For future work, we recommend enhancing the classification and mapping field to the utilization of drones in agriculture since drones can reach a vast area that is very fit for fertilizing, spraying, or even capturing crop images with live cameras to identify leaf disease
IMPLEMENTATION OF JOHNSON\u27S SHORTEST PATH ALGORITHM FOR ROUTE DISCOVERY MECHANISM ON SOFTWARE DEFINED NETWORK
Full text linkSoftware Defined Network is a network architecture with a new paradigm which consists of a control plane that is placed separately from the data plane. All forms of computer network behavior are controlled by the control plane. Meanwhile the data plane consisting of a router or switch becomes a device for packet forwarding. With a centralized control plane model, SDN is very vulnerable to congestion because of the one-to-many communication model. There are several mechanisms for congestion control on SDNs, one of which is modifying packets by reducing the size of packets sent. But this is considered less effective because the time required will be longer because the number of packets sent is less. This requires that network administrators must be able to configure a network with certain routing protocols and algorithms. Johnson\u27s algorithm is used in determining the route for packet forwarding, with the nature of the all-pair shortest path that can be applied to SDN to determine through which route the packet will be forwarded by comparing all nodes that are on the network. The results of the Johnson algorithm\u27s latency and throughput with the comparison algorithm show good results and the comparison of the Johnson algorithm\u27s trial results is still superior. The response time results of the Johnson algorithm when first performing a route search are faster than the conventional OSPF algorithm due to the characteristics of the all pair shortest path algorithm which determines the shortest route by comparing all pairs of nodes on the network
MAPPING POTENTIAL ATTACKERS AGAINST NETWORK SECURITY USING LOCATION AWARE REACHABILITY QUERIES ON GEO SOCIAL DATA
Full text linkAttacks on network security can happen anywhere. Using Geo-Social Networks (GSN), i.e., a graph that combines social network data and spatial information, we can find the potential attackers based on the given location. In answering the graph-based problems, Reachability Queries are utilized. It verifies the reachability between two nodes in the graph. This paper addresses a problem defined as follows: Given a geo-social graph and a location area as a query point, we map potential attackers against network security using location-aware reachability queries. We employ the concepts of Reachability Minimum Bounding Rectangle (RMBR) and graph traversal algorithm, i.e., Depth-First Search (DFS), to answer the location-aware reachability queries. There are two kinds of the proposed solution, i.e., (1) RMBR-based solution map potential attackers by looking for intersecting RMBR values, and (2) Graph traversal-based solution map potential attackers by traversing the graph. We evaluate the performance of both proposed solutions using synthetic datasets. Based on the experimental result, the RMBR-based solution has much lower execution time and memory usage than the graph traversal-based solution
INSTRUMENTATION-BASED MONITORING TECHNIQUES SURVEY ON HOST, PLATFORM, AND SERVICE LEVEL IN MICROSERVICE ARCHITECTURE
Full text linkMicroservice is an application architecture that separates one big application into smaller ones. The architecture simplifies development, deployment, and management process. However, the architecture is quite complex thus the monitoring process becomes much more challenging. Classifications for the instrumentations that are used in the monitoring process is needed to achieve better practicality for the administrators. We surveyed the monitoring technique classification method in microservice architecture. The method is divided into three levels. They are host level, platform level, and service level. In this paper, we present the latest instruments that are being used in the monitoring process in each level. Correlation between the goals, needs, and stakeholder is also presented
IMPROVED DEEP LEARNING ARCHITECTURE WITH BATCH NORMALIZATION FOR EEG SIGNAL PROCESSING
Full text linkDeep learning is commonly used to solve problems such as biomedical problems and many other problems. The most common architecture used to solve those problems is Convolutional Neural Network (CNN) architecture. However, CNN may be prone to overfitting, and the convergence may be slow. One of the methods to overcome the overfitting is batch normalization (BN). BN is commonly used after the convolutional layer. In this study, we proposed a further usage of BN in CNN architecture. BN is not only used after the convolutional layer but also used after the fully connected layer. The proposed architecture is tested to detect types of seizures based on EEG signals. The data used are several sessions of recording signals from many patients. Each recording session produces a recorded EEG signal. EEG signal in each session is first passed through a bandpass filter. Then 26 relevant channels are taken, cut every 2 seconds to be labeled the type of epileptic seizure. The truncated signal is concatenated with the truncated signal from other sessions, divided into two datasets, a large dataset, and a small dataset. Each dataset has four types of seizures. Each dataset is equalized using the undersampling technique. Each dataset is then divided into test and train data to be tested using the proposed architecture. The results show the proposed architecture achieves 46.54% accuracy for the large dataset and 93.33% accuracy for the small dataset. In future studies, the batch normalization parameter will be further investigated to reduce overfitting
MODIFIED LOCAL TERNARY PATTERN WITH CONVOLUTIONAL NEURAL NETWORK FOR FACE EXPRESSION RECOGNITION
Full text linkFacial expression recognition (FER) on images with illumination variation and noises is a challenging problem in the computer vision field. We solve this using deep learning approaches that have been successfully applied in various fields, especially in uncontrolled input conditions. We apply a sequence of processes including face detection, normalization, augmentation, and texture representation, to develop FER based on Convolutional Neural Network (CNN). The combination of TanTriggs normalization technique and Adaptive Gaussian Transformation Method is used to reduce light variation. The number of images is augmented using a geometric augmentation technique to prevent overfitting due to lack of training data. We propose a representation of Modified Local Ternary Pattern (Modified LTP) texture image that is more discriminating and less sensitive to noise by combining the upper and lower parts of the original LTP using the logical AND operation followed by average calculation. The Modified LTP texture images are then used to train a CNN-based classification model. Experiments on the KDEF dataset show that the proposed approach provides a promising result with an accuracy of 81.15%
IMPERSONATION METHOD ON AUTHORIZATION SERVER USING CLIENT-INITIATED BACK-CHANNEL AUTHENTICATION PROTOCOL
Full text linkThere is an impersonation (login as) feature in several applications that can be used by system administrators who have special privileges. This feature can be utilized by development and maintenance teams that have administrator rights to reproduce errors or bugs, to check specific features in applications according to the specific users’ login sessions. Beside its benefits, there is a security vulnerability that allows administrators to abuse the rights. They can access users’ private data or execute some activities inside the system without account or resource owners’ consents.This research proposes an impersonation method on authorization server using Client-Initiated Back-channel Authentication (CIBA) protocol. This method prevents impersonation without account or resource owners’ consent. The application will ask users’ authentication and permission via authentication device possessed by resource owners before the administrator performs impersonation. By utilizing authentication device, the impersonation feature should be preceded by users’ consent and there is no direct interaction needed between the administrator and resource owners to prove the users’ identities. The result shows that the implementation of CIBA protocol can be used to complement the impersonation method and can also run on the authorization server that uses OAuth 2.0 and OpenID Connect 1.0 protocols. The system testing is done by adopting FAPI CIBA conformance testing