• Journal of the Institute of Convergence Signal Processing
• /
• v.22 no.3
• /
• pp.99-103
• /
• 2021

Recently, convolutional neural networks (CNNs) have been widely utilized due to their outstanding performance in various computer vision fields. However, due to their computational-intensive and high memory requirements, it is difficult to deploy CNNs on hardware platforms that have limited resources, such as mobile devices and IoT devices. To address these limitations, a neural network compression research is underway to reduce the size of neural networks while maintaining their performance. This paper proposes a CNN compression technique that dynamically adjusts the thresholds of pruning, one of the neural network compression techniques. Unlike the conventional pruning that experimentally or heuristically sets the thresholds that determine the weights to be pruned, the proposed technique can dynamically find the optimal thresholds that prevent accuracy degradation and output the light-weight neural network in less time. To validate the performance of the proposed technique, the LeNet was trained using the MNIST dataset and the light-weight LeNet could be automatically obtained 1.3 to 3 times faster without loss of accuracy.

• KIPS Transactions on Software and Data Engineering
• /
• v.4 no.9
• /
• pp.347-356
• /
• 2015

Due to the rapid development of IoT(Internet of Things) technology, traditional taxis are connected through dispatchers and location systems. Typically, modern taxis have embedded with GPS(Global Positioning System), which aims for obtaining the route information. By analyzing the frequency of taxi trip events, we can find the frequent route for a given query time. However, a scalability problem would occur when we convert the raw location data of taxi trip events into the analyzed frequency information due to the volume of location data. For this problem, we propose a NoSQL based top-K query system for taxi trip events. First, we analyze raw taxi trip events and extract frequencies of all routes. Then, we store the frequency information into hash-based index structure of MongoDB which is a document-oriented NoSQL database. Efficient top-K query processing for frequent route is done with the top of the MongoDB. We validate the efficiency of our algorithms by using real taxi trip events of New York City.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.3
• /
• pp.70-77
• /
• 2017

Recently, Internet of Things (IoT) and deep learning techniques have affected video surveillance systems in various ways. The surveillance features that perform detection, tracking, and classification of specific objects in Closed Circuit Television (CCTV) video are becoming more intelligent. This paper presents real-time algorithm that can run in a PC environment using only a low power CPU. Traditional tracking algorithms combine background modeling using the Gaussian Mixture Model (GMM), Hungarian algorithm, and a Kalman filter; they have relatively low complexity but high detection errors. To supplement this, deep learning technology was used, which can be trained from a large amounts of data. In particular, an SRGB(Sequential RGB)-3 Layer CNN was used on tracked objects to emphasize the features of moving people. Performance evaluation comparing the proposed algorithm with existing ones using HOG and SVM showed move-in and move-out error rate reductions by 7.6 % and 9.0 %, respectively.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.17 no.8
• /
• pp.311-317
• /
• 2016

Many services have been developed that are based on smart devices, and security between devices is emphasized. A beacon on the current IoT(Internet of Things) services has been utilized in the commercial field and is being applied to the services of the home IoT. On the other hand, the beacon is weak to security using Bluetooth-based services. Therefore, it is important to strengthen the security of the beacon. This paper proposes a dual security technique that can enhance the security of beacon-based services. The dual security architecture and security process is proposed based on beacon and authentication service. In addition, mobile application was developed and validated based on the beacon for proving the suitability of the proposed technique. The experimental method for verification are the authentication failure case, such as 1st authentication fail, and authentication success case, such as 1st authentication success and 2nd authentication success. The components of the verification experiments consists of two beacons (matched with Beacon ID, mismatched with Beacon ID), one mobile device and authentication application. This was tested to verify the compatibility of the dual security architecture and 1st/2nd authentication process.

• Journal of KIISE
• /
• v.42 no.12
• /
• pp.1474-1479
• /
• 2015

With mobile devices showing marked improvement in performance in the age of the Internet of Things (IoT), there is demand for rapid processing of the extensive amount of multimedia big data. However, because research on image searching is focused mainly on increasing accuracy despite environmental changes, the development of fast processing of high-resolution multimedia data queries is slow and inefficient. Hence, we suggest a new distributed image search algorithm that ensures both high accuracy and rapid response by using feature extraction of distributed images based on MapReduce, and solves the problem of memory scalability based on BIRCH indexing. In addition, we conducted an experiment on the accuracy, processing time, and scalability of this algorithm to confirm its excellent performance.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.30 no.6
• /
• pp.1279-1289
• /
• 2020

To securely protect users' sensitive information and national secrets, the importance of cryptographic modules has been emphasized. Currently, many companies and national organizations are actively using cryptographic modules. In Korea, To ensure the security of these cryptographic modules, the cryptographic module has been verified through the Korea Certificate Module Validation Program(KCMVP). Most of the domestic cryptographic modules are CPU-based software (S/W). However, CPU-based cryptographic modules are difficult to use in servers that need to process large amounts of data. In this paper, we propose an S/W cryptographic module that provides a high-speed operation using GPU. We describe the configuration and operation of the S/W cryptographic module using GPU and present the changes in the cryptographic module security requirements by using GPU. In addition, we present the performance improvement compared to the existing CPU S/W cryptographic module. The results of this paper can be used for cryptographic modules that provide cryptography in servers that manage IoT (Internet of Things) or provide cloud computing.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.15 no.1
• /
• pp.205-210
• /
• 2020

Children should be provided with the right to be protected by adults, related organizations, or countries because they lack power, situation recognition ability, and situation judgment ability compared to adults. Also, children's accidents and accidents caused by neglect of management occur every year. Currently, there are about 3 to 20 children managed by one teacher in a daycare center, and it is very difficult for teachers to care for all children as many children have to be managed. Especially, when outdoor activities are conducted in open space, children's activities are expanded compared to the in-house, and children who do not follow the control of the in-house teacher may occur, so there is a limit to the control and management of children depending on the viewpoint of teachers. In this study, we designed and implemented IoT terminal and system to provide safety to children and to provide convenience to guardians and teachers by systemizing the location or interval between children and in-service teachers using portable devices that can communicate with each other. It is hoped that this will contribute to the safety of infants.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.13 no.1
• /
• pp.221-228
• /
• 2018

Big data is an important keyword in World's Fourth Industrial Revolution in public and private division including IoT(Internet of Things), AI(Artificial Intelligence) and Cloud system in the fields of science, technology, industry and society. Big data based on services are available in various fields such as transportation, weather, medical care, and marketing. In particular, in the field of sports, various types of bio-signals can be collected and managed by the appearance of a wearable device that can measure vital signs in training or rehabilitation for daily life rather than a hospital or a rehabilitation center. However, research on big data with vital signs from wearable devices for training and rehabilitation for baseball players have not yet been stimulated. Therefore, in this paper, we propose a system for baseball infield and outfield players, especially which can store and analyze the momentum measurement vital signals based on big data.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.10a
• /
• pp.1013-1014
• /
• 2015

Regardless of the strict enforcement of prevention education in accordance with the school facility management standard, safety accidents that lead to human and physical damages occur in current educational facilities because of teenagers with very low sense of responsibility and insensitivity toward fire and facility safety. To ensure educational facility safety, technology that will enable a fast work process and easy confirmation of electronic blueprints and related documents about the educational facility through smart devices at the site by various means is needed. This paper proposes a system design linked to the National Education Information System (NEIS) that uses the document conversion function, high efficiency resolution, and Internet of Things (IoT) to inspect and control the educational facility in the event of a safety accident through the Educational Facility Resource Management System (EFRMS) that manages the electronic blueprints, and various educational facility documents through various smart devices.

• Convergence Security Journal
• /
• v.19 no.1
• /
• pp.121-126
• /
• 2019

The big data utilization category in the sports field was mainly focused on the big data analysis to improve the competence of the athlete and the performance. Since then, 'big data technology' which collect and analyze more detailed and diverse data through the application of ICT technology such as IoT and AI has been applied. The use of big data of sports contents in future has value and possibility in the smart environment, but it is necessary to overcome the shortage and limitation of platform to manage and share sports contents. In order to solve such problems, it is important to change the perception of the companies or providers that provide sports contents and cultivate and secure professional personnel capable of providing sports contents. Also, it is necessary to implement policies to systematically manage and utilize big data poured from sports contents.