JOIV : International Journal on Informatics Visualization
Not a member yet
786 research outputs found
Sort by
Design of a Low-area Digit Recognition Accelerator Using MNIST Database
Get PDFDeep neural networks, which is a field of artificial intelligence, have been used in various fields. Deep learning is processed on high-performance GPUs or TPUs. It requires high cost as much as its good performance. Recently, as the demand for edge computing increases, many studies have been conducted to perform complex deep learning operations in a low-computing processor. Among them, a typical study is to lighten the deep learning network. In this paper, we propose a handwritten digit recognition hardware accelerator suitable for edge computing using MNIST database. After setting the correction rate for MNIST to 94% and performing network lighting processes, a hardware structure that can reduce the area of hardware and minimize memory access is proposed. Basically, the network is set as a two-layer fully connected network. The network is modeled with Python and lighten while checking the performance. Network parameters, weighs and biases, are quantized. The pixel number and bit number of MNIST input data are also reduced. The number of MAC units and the processing order of the hardware accelerator are determined so that there is no not used MACs while performing the MAC operations in parallel. It is designed with Verilog HDL and its functions are checked in Modelsim. And then it is implemented in Xilinx Zynq ZC-702 to verify the operations. The designed number recognition accelerator is expected to be widely used in edge devices by reducing the area and memory access
Development on Deaf Support Application Based on Daily Sound Classification Using Image-based Deep Learning
Get PDFAccording to statistics, the number of hearing-impaired persons among the disabled in Korea accounts for 27% of all persons with disabilities. However, there is insufficient support for the deaf and hard of hearing's protective devices and life aids compared to the large number. In particular, the hearing impaired misses much information obtained through sound and causes inconvenience in daily life. Therefore, in this paper, we propose a method to relieve the discomfort in the daily life of the hearing impaired. It analyzes sounds that can occur frequently and must be recognized in daily life and guide them to the hearing impaired through applications and vibration bracelets. Sound analysis was learned by using deep learning by converting sounds that often occur in daily life into the Mel-Spectrogram. The sound that actually occurs is recorded through the application, and then it is identified based on the learning result. According to the identification result, predefined alarms and vibrations are provided differently so that the hearing impaired can easily recognize it. As a result of the recognition of the four major sounds occurring in real life in the experiment, the performance showed an average of 85% and an average of 80% of the classification rate for mixed sounds. It was confirmed that the proposed method can be applied to real-life through experiments. Through the proposed method, the quality of life can be improved by allowing the hearing impaired to recognize and respond to sounds that are essential in daily life
Immersive Applications in Museums: An Analysis of the Use of XR Technologies and the Provided Functionality Based on Systematic Literature Review
Get PDFImmersive technologies (Virtual, Augmented, and Mixed Reality) are widely used in cultural heritage for communication, enhancing the visiting experience, and improving learning and understanding. Immersive technologies have found their way into museums and other cultural spaces in various forms and shapes. This work aims to recognize the main forms of immersive technologies and applications in museums and other cultural spaces and provide information on the employed methods, technologies, equipment, and software solutions by conducting a systematic literature review aligned with the PRISMA guidelines. The analyzed literature was collected through a focused search in scientific databases (Scopus, ACM, and IEEE). The relevance to the subject was assessed based on the main technological focus (VR/AR/MR or XR) and the employed technologies. Methods and approaches for realizing their applications were studied and discussed. Thirteen articles were found to meet the selection criteria, of which two focus on VR, six are on AR, two are on Audio-AR, and three are on MR. The results showed that Augmented Reality solutions are preferred for on-site use; Mixed Reality applications started to emerge as Mixed Reality hardware technology became available and Virtual Reality despite being criticized for isolating visitors. The findings cover the existing gap in recent literature and can reveal a set of good practices and innovative ideas for future applications
Academic Document Authentication using Elliptic Curve Digital Signature Algorithm and QR Code
Get PDFPaper-based documents or printed documents such as recommendation letters, academic transcripts, and diplomas are prone to forgery. Several methods have been used to protect them, such as watermarking, security holograms, or using paper with specific security features. This paper presents a document authentication system that utilizes QR code and ECDSA as the digital signature algorithm to protect this kind of document from counterfeiting. A digital signature is a well-known technique in modern cryptography used for providing data integrity and authentication. The idea proposed herein is to put a QR code in the printed documents where the QR code includes a digital signature. The signature can later be authenticated using the proposed system by uploading the document for authentication or scanning the document's QR code. The proposed system is particularly developed for digital signature generation and verification of students' final project approval documents as the case study. In traditional settings, the approval form is typically signed directly by the student's advisor dan co-advisor using handwritten signatures. However, using the conventional handwritten signature, the signature on the approval form can be falsified. Therefore, a digital signature generation and verification system is implemented herein to avoid handwritten signature falsification. The advisors can use this system to sign the approval form using a digital signature instead of a handwritten one. The signature is stored in a QR code and is generated using ECDSA with SHA-256 as the hash function. The proposed system is evaluated using documents (i.e., approval forms) with genuine and forged QR codes. The evaluation results showed that the system could verify the authenticity of the approval forms, which contain genuine QR codes. The approval forms that contained forged QR codes were correctly identified
Malware Authorship Attribution Model using Runtime Modules based on Automated Analysis
Get PDFMalware authorship attribution is a research field that identifies the author of malware by extracting and analyzing features that relate the authors from the source code or binary code of malware. Currently, it is being used as one of the detection techniques based on malware forensics or identifying patterns of continuous attacks such as APT attacks. The analysis methods to identify the author are as follows. One is a source code-based analysis method that extracts features from the source code, and the other is a binary-based analysis method that extracts features from the binary. However, to handle the modularization and the increasing amount of malicious code with these methods, both time and manpower are insufficient to figure out the characteristics of the malware. Therefore, we propose the model for malware authorship attribution by rapidly extracting and analyzing features using automated analysis. Automated analysis uses a tool and can be analyzed through a file of malware and the specific hash values without experts. Furthermore, it is the fastest to figure out among other malware analysis methods. We have experimented by applying various machine learning classification algorithms to six malware author groups, and Runtime Modules and Kernel32.dll API extracted from the automated analysis were selected as features for author identification. The result shows more high accuracy than the previous studies. By using the automated analysis, it extracts features of malware faster than source code and binary-based analysis methods
Exploration of The Impact of Kernel Size for YOLOv5-based Object Detection on Quadcopter
Get PDFDrones or quadcopters have been widely used in various fields based on deep learning, especially object detection. However, drone vision characteristics such as occlusion and small objects are still being explored for performance in terms of accuracy and speed detection. The YOLO architecture is very commonly used for cases requiring high-speed detection. To overcome the limitations of drone vision, in this paper, we explore the size of the YOLOv5s backbone kernel in the shallowest convolutional layer to achieve better performance. The kernel is a filter that has a main role in the feature map, and it defines the size of the convolution matrix, and the resulting features in the shallowest convolutional layer are more representative of the case of object detection and recognition. The techniques can be divided into three major categories: (1) data preprocessing, which involves augmentation and normalization of the data, (2) kernel size exploration in the shallowest convolutional layer of the YOLOv5s, and (3) model implementation in the real environment using the quadcopter. The dataset consisted of four classes representing dragon fruit, snake fruit, banana, and pineapple, with a total of 8000 data. Exploration results with kernel size give promising results. Kernel sizes 5 and 7 give an mAP of 0.988. Through these results, modification of the kernel size provides an opportunity for more in-depth investigations, such as with the epoch parameter, padding scheme, and other optimization techniques
Application of Gray Scale Matrix Technique for Identification of Lombok Songket Patterns Based on Backpropagation Learning
Get PDFSongket is a woven fabric created by prying the threads and adding more weft to create an embossed decorative pattern on a cotton or silk thread woven background. While songket from many places share similar motifs, when examined closely, the motifs of songket from various regions differ, one of which is in the Province of West Nusa Tenggara, namely Lombok Island. To assist the public in recognizing the many varieties of Lombok songket motifs, the researchers used digital image processing technology, including pattern recognition, to distinguish the distinctive patterns of Lombok songket. The Gray Level Co-occurrence Matrix (GLCM) technique and Backpropagation Neural Networks are used to build a pattern identification system to analyze the Lombok songket theme. Before beginning the feature extraction process, the RGB color image has converted to grayscale (grayscale), which is resized. Simultaneously, a Backpropagation Neural Network is employed to classify Lombok songket theme variations. This study used songket motif photos consisting of a sample of 15 songket motifs with the same color theme that was captured eight times, four of which were used as training data and kept in the database. Four additional photos were utilized as test data or data from sources other than the database. When the system’s ability to recognize the pattern of Lombok songket motifs is tested, the maximum average recognition percentage at a 0° angle is 88.33 percent. In comparison, the lowest average recognition percentage at a 90° angle is 68.33 percent
A Univariate Extreme Value Analysis and Change Point Detection of Monthly Discharge in Kali Kupang, Central Java, Indonesia
Get PDFKali Kupang plays an important role in the life of the people of Pekalongan and its surrounding areas. However, until recently, not many hydrological studies have been carried out in this area. This study presents how Extreme Value Analysis (EVA) can predict future extreme hydrological events and how a dynamic-programming-based change point detection algorithm can detect the abrupt transition in discharge events variability. Using the annual block maxima, we can predict the upper extreme discharge probability from the generalized extreme value distribution (GEVD) that best fits the data by using the Markov Chain Monte Carlo (MCMC) algorithm as a distribution fitting method. Metropolis-Hasting (MH) algorithm with 500 walkers and 2,500 samples for each walker is used to generate random samples from the prior distribution. As a result, this discharge data can be categorized as a Gumbel distribution (  = 6.818,  = 3.456, and  = 0). The recurrence intervals (RI) for this discharge data can be calculated through this distribution. The changepoint location of the annual standard deviation of this discharge data in the mid-1990s is detected by using the pruned exact linear time (PELT) algorithm. Despite some shortcomings, this study can pave the way for using data-driven algorithms, along with more traditional numerical and descriptive approaches, to analyze hydrological time-series data in Indonesia. This is crucial, considering an increasing number of hydro climatological disasters in the future as a consequence of global climate change
Enhancing Code Similarity with Augmented Data Filtering and Ensemble Strategies
Get PDFAlthough COVID-19 has severely affected the global economy, information technology (IT) employees managed to perform most of their work from home. Telecommuting and remote work have promoted a demand for IT services in various market sectors, including retail, entertainment, education, and healthcare. Consequently, computer and information experts are also in demand. However, producing IT, experts is difficult during a pandemic owing to limitations, such as the reduced enrollment of international students. Therefore, researching increasing software productivity is essential; this study proposes a code similarity determination model that utilizes augmented data filtering and ensemble strategies. This algorithm is the first automated development system for increasing software productivity that addresses the current situation—a worldwide shortage of software dramatically improves performance in various downstream natural language processing tasks (NLP). Unlike general-purpose pre-trained language models (PLMs), CodeBERT and GraphCodeBERT are PLMs that have learned both natural and programming languages. Hence, they are suitable as code similarity determination models. The data filtering process consists of three steps: (1) deduplication of data, (2) deletion of intersection, and (3) an exhaustive search. The best mating (BM) 25 and length normalization of BM25 (BM25L) algorithms were used to construct positive and negative pairs. The performance of the model was evaluated using the 5-fold cross-validation ensemble technique. Experiments demonstrate the effectiveness of the proposed method quantitatively. Moreover, we expect this method to be optimal for increasing software productivity in various NLP tasks
An Improved Approach of Iris Biometric Authentication Performance and Security with Cryptography and Error Correction Codes
Get PDFOne of the most challenging parts of integrating biometrics and cryptography is the intra variation in acquired identifiers between the same users. Due to noise in the environment or different devices, features of the iris may differ when it is acquired at different time periods. This research focuses on improving the performance of iris biometric authentication and encrypting the binary code generated from the acquired identifiers. The proposed biometric authentication system incorporates the concepts of non-repudiation and privacy. These concepts are critical to the success of a biometric authentication system. Iris was chosen as the biometric identifier due to its characteristics of high accuracy and the permanent presence throughout an individual’s lifetime. This study seeks to find a method of reducing the noise and error associated with the nature of dissimilarity acquired by each biometric acquisition. In our method, we used Reed Solomon error-correction codes to reduce dissimilarities and noise in iris data. The code is a block-based error correcting code that can be easily decoded and has excellent burst correction capabilities. Two different distance metric measurement functions were used to measure the accuracy of the iris pattern matching identification process, which are Hamming distance and weighted Euclidean distance. The experiments were conducted with the CASIA 1.0 iris database. The results showed that the False Acceptance Rate is 0%, the False Rejection Rate is 1.54%, and the Total Success Rate is 98.46%. The proposed approach appears to be more secure, as it is able to provide a low rate of false rejections and false acceptances