• Journal of the Korean Society for Railway
• /
• v.19 no.4
• /
• pp.445-455
• /
• 2016

A goal of the Real-time railway safety supervision system is to improve the safety oversight efficiency and to prevent accidents by integrating existing distributed monitoring systems, train, signal, power and facilities. So, the system require better performance regarding real-time processing based on big data. The disk-based database that is used in existing railway control systems has a problem with real-time processing; memory-based databases haves a limitation in terms of big-data processing; and time series databases haves a limitation in terms of real-time processing. So, we need a new database architecture for simultaneous real-time processing based on big data. In this study, we review the existing railway monitoring systems and propose a new database architecture for a real-time railway safety supervision system.

• Journal of Information Processing Systems
• /
• v.14 no.6
• /
• pp.1464-1479
• /
• 2018

In face recognition, sometimes the number of available training samples for single category is insufficient. Therefore, the performances of models trained by convolutional neural network are not ideal. The small sample face recognition algorithm based on novel Siamese network is proposed in this paper, which doesn't need rich samples for training. The algorithm designs and realizes a new Siamese network model, SiameseFacel, which uses pairs of face images as inputs and maps them to target space so that the $L_2$ norm distance in target space can represent the semantic distance in input space. The mapping is represented by the neural network in supervised learning. Moreover, a more lightweight Siamese network model, SiameseFace2, is designed to reduce the network parameters without losing accuracy. We also present a new method to generate training data and expand the number of training samples for single category in AR and labeled faces in the wild (LFW) datasets, which improves the recognition accuracy of the models. Four loss functions are adopted to carry out experiments on AR and LFW datasets. The results show that the contrastive loss function combined with new Siamese network model in this paper can effectively improve the accuracy of face recognition.

• Advanced Industrial SCIence
• /
• v.3 no.2
• /
• pp.17-22
• /
• 2024

Recently, as the ICT field has been used in various environments, it has become possible to analyze pests by crops, use robots when harvesting crops, and predict by big data by utilizing ICT technologies in a sustainable agricultural environment. However, in a sustainable agricultural environment, efforts to solve resource depletion, agricultural population decline, poverty increase, and environmental destruction are constantly being demanded. This paper proposes an artificial intelligence-based big data processing analysis method to reduce the production cost and increase the efficiency of crops based on a sustainable agricultural environment. The proposed technique strengthens the security and reliability of data by processing big data of crops combined with AI, and enables better decision-making and business value extraction. It can lead to innovative changes in various industries and fields and promote the development of data-oriented business models. During the experiment, the proposed technique gave an accurate answer to only a small amount of data, and at a farm site where it is difficult to tag the correct answer one by one, the performance similar to that of learning with a large amount of correct answer data (with an error rate within 0.05) was found.

• Annual Conference of KIPS
• /
• 2023.11a
• /
• pp.963-964
• /
• 2023

추천 시스템은 사용자의 선호도를 정확히 파악하는 것이 중요하다. 이를 위해 사용자 데이터를 분석하여 추천을 제공하는 협업 필터링 알고리즘을 활용한다. 하지만 상품의 종류와 고객 수가 많아짐에 따라 사용자 선호도 정확도가 떨어지는 문제점이 있다. 이 문제를 해결하기 위해 제안된 방법은 모델 기반 협업 필터링이며, 이는 고객과 사용자의 정보를 직접적으로 추천하는 대신 모델을 학습시키는데 활용된다. 이에 논문은 추천시스템에서 자주 사용되는 모델 협업 필터링 기반 SVD 모델을 학습 전에 하이퍼파라미터를 조절하여 모델에 추정 정확도 값인 RMSE를 측정한다.

• Annual Conference of KIPS
• /
• 2023.11a
• /
• pp.1014-1015
• /
• 2023

많은 장애아 부모들은 양육에 대한 스트레스, 미래에 대한 걱정으로 심리적으로 상당한 중압감을 느낀다. 이에 비해 매년 증가하는 장애인 수에 비해 장애아 부모 및 가족의 심리적·정신적 문제를 해결하기 위한 프로그램이 부족하다.[1] 이를 해결하고자 본 논문에서는 감정분석 소통 플랫폼을 제안한다. 제안하는 플랫폼은 KoBERT 모델을 fine-tunning 하여 사용자의 일기 속 감정을 분석하여 장애아를 둔 부모 및 가족 간의 소통을 돕는다. 성능평가는 제안하는 플랫폼의 주요 기능인 KoBERT 기반 감정분석의 성능을 확인하기위해 텍스트 분류 모델로 널리 사용되고 있는 LSTM, Bi-LSTM, GRU 모델 별 성능지표들과 비교 분석한다. 성능 평가결과 KoBERT 의 정확도가 다른 분류군의 정확도보다 평균 31.4% 높은 성능을 보였고, 이 외의 지표에서도 비교적 높은 성능을 기록했다.

• Journal of Industrial Technology
• /
• v.36
• /
• pp.89-92
• /
• 2016

In this paper, we propose an efficient anonymity algorithm for personal information protections in big data systems. Firstly, we briefly introduce fundamental algorithms of k-anonymity, l-diversity, t-closeness. And then we propose an anonymity algorithm using controlling the size of anonymity groups as well as exchanging the data tuple between anonymity groups. Finally, we demonstrate an example on which proposed algorithm applied. The proposed scheme gave an efficient and simple algorithms for the processing of a big amount of data.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.4
• /
• pp.1146-1165
• /
• 2022

A huge amount of data in the form of videos and images is being produced owning to advancements in sensor technology. Use of low performance commodity hardware coupled with resource heavy image processing and analyzing approaches to infer and extract actionable insights from this data poses a bottleneck for timely decision making. Current approach of GPU assisted and cloud-based architecture video analysis techniques give significant performance gain, but its usage is constrained by financial considerations and extremely complex architecture level details. In this paper we propose a data pipeline system that uses open-source tools such as Apache Spark, Kafka and OpenCV running over commodity hardware for video stream processing and image processing in a distributed environment. Experimental results show that our proposed approach eliminates the need of GPU based hardware and cloud computing infrastructure to achieve efficient video steam processing for face detection with increased throughput, scalability and better performance.

• Journal of the Korea Society of Computer and Information
• /
• v.23 no.3
• /
• pp.41-48
• /
• 2018

The recent sudden increase of big data has characteristics such as continuous generation of data, large amount, and unstructured format. The existing relational database technologies are inadequate to handle such big data due to the limited processing speed and the significant storage expansion cost. Current implemented solutions are mainly based on relational database that are no longer adapted to these data volume. NoSQL solutions allow us to consider new approaches for data warehousing, especially from the multidimensional data management point of view. In this paper, we develop and propose the integrated design methodology based on MongoDB for big data applications. The proposed methodology is more scalable than the existing methodology, so it is easy to handle big data.

• Ocean Systems Engineering
• /
• v.9 no.1
• /
• pp.79-95
• /
• 2019

The material analysis of an offshore structure is generally conducted in the contract design phase for the price quotation of a new offshore project. This analysis is conducted manually by an engineer, which is time-consuming and can lead to inaccurate results, because the data size from previous projects is too large, and there are so many materials to consider. In this study, the piping materials in an offshore structure are analyzed for contract design using a big data framework. The big data technologies used include HDFS (Hadoop Distributed File System) for data saving, Hive and HBase for the database to handle the saved data, Spark and Kylin for data processing, and Zeppelin for user interface and visualization. The analyzed results show that the proposed big data framework can reduce the efforts put toward contract design in the estimation of the piping material cost.

• Journal of Information Processing Systems
• /
• v.13 no.6
• /
• pp.1575-1584
• /
• 2017

As interest in big data has increased recently, NoSQL, a solution for storing and processing big data, is getting attention. NoSQL supports high speed, high availability, and high scalability, but is limited in areas where data integrity is important because it does not support multiple row transactions. To overcome these drawbacks, many studies are underway to support multiple row transactions in NoSQL. However, existing studies have a disadvantage that the number of transactions that can be processed per unit of time is low and performance is degraded. Therefore, in this paper, we design and implement a multi-row transaction system for data integrity in big data environment based on HBase, a column-based NoSQL which is widely used recently. The multi-row transaction system efficiently performs multi-row transactions by adding columns to manage transaction information for every user table. In addition, it controls the execution, collision, and recovery of multiple row transactions through the transaction manager, and it communicates with HBase through the communication manager so that it can exchange information necessary for multiple row transactions. Finally, we performed a comparative performance evaluation with HAcid and Haeinsa, and verified the superiority of the multirow transaction system developed in this paper.