• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.10a
• /
• pp.1019-1022
• /
• 2019

최근의 영상 처리 분야는 딥러닝 기법들의 성능이 입증됨에 따라 다양한 분야에서 이와 같은 기법들을 활용해 영상에 대한 분류, 분석, 검출 등을 수행하려는 시도가 활발하다. 그중에서도 의료 진단 보조 역할을 할 수 있는 의료 영상 분석 소프트웨어에 대한 기대가 증가하고 있는데, 본 연구에서는 캡슐내시경 영상에 주목하였다. 캡슐내시경은 주로 소장 촬영을 목표로 하며 식도부터 대장까지 약 8~10시간 동안 촬영된다. 이로 인해 CT, MR, X-ray와 같은 다른 의료 영상과 다르게 하나의 데이터 셋이 10~15만 장의 이미지를 갖는다. 일반적으로 캡슐내시경 영상을 판독하는 순서는 위장관 교차점(Z-Line, 유문판, 회맹판)을 기준으로 위장관 랜드마크(식도, 위, 소장, 대장)를 구분한 뒤, 각 랜드마크 별로 병변 정보를 찾아내는 방식이다. 그러나 워낙 방대한 영상 데이터를 가지기 때문에 의사 혹은 의료 전문가가 영상을 판독하는데 많은 시간과 노력이 소모되고 있다. 본 논문의 목적은 캡슐내시경 영상의 판독에서 모든 환자에 대해 공통으로 수행되고, 판독하는 데 많은 시간을 차지하는 위장관 랜드마크를 찾는 것에 있다. 이를 위해, 위장관 랜드마크를 식별할 수 있는 CNN 학습 모델을 설계하였으며, 더욱 효과적인 학습을 위해 전처리 과정으로 학습에 방해가 되는 학습 노이즈 영상들을 제거하고 위장관 랜드마크 별 특징 분석을 진행하였다. 총 8명의 환자 데이터를 가지고 학습된 모델에 대해 평가 및 검증을 진행하였는데, 무작위로 환자 데이터를 샘플링하여 학습한 모델을 평가한 결과, 평균 정확도가 95% 가 확인되었으며 개별 환자별로 교차 검증 방식을 진행한 결과 평균 정확도 67% 가 확인되었다.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.32 no.5
• /
• pp.945-954
• /
• 2022

Billions of malicious codes are detected every year, of which only 0.01% are new types of malware. In this situation, an effective malware type classification tool is needed, but previous studies have limitations in quickly analyzing a large amount of malicious code because it requires a complex and massive amount of data pre-processing. To solve this problem, this paper proposes a method to classify the types of malicious code based on the similarity hash without complex data preprocessing. This approach trains the XGBoost model based on the similarity hash information of the malware. To evaluate this approach, we used the BIG-15 dataset, which is widely used in the field of malware classification. As a result, the malicious code was classified with an accuracy of 98.9% also, identified 3,432 benign files with 100% accuracy. This result is superior to most recent studies using complex preprocessing and deep learning models. Therefore, it is expected that more efficient malware classification is possible using the proposed approach.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.7
• /
• pp.57-65
• /
• 2023

Various studies to enhance the accuracy of rice production forecasting are focused on improving the accuracy of the models. In contrast, there is a relative lack of research regarding the data itself, which the prediction models are applied to. When applying the same dependent variable and prediction model to two different sets of rice production data composed of distinct features, discrepancies in results can occur. It is challenging to determine which dataset yields superior results under such circumstances. To address this issue, by identifying potential influential features within the data before applying the prediction model and centering the modeling around these, it is possible to achieve stable prediction results regardless of the composition of the data. In this study, we propose a method to adjust the composition of the data's features in order to select optimal base variables, aiding in achieving stable and consistent predictions for rice production. This method makes use of the Korea Meteorological Administration's ASOS data. The findings of this study are expected to make a substantial contribution towards enhancing the utility of performance evaluations in future research endeavors.

• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.408-415
• /
• 2022

In this paper, we propose a 3D point cloud reconstruction technique from 2D images using efficient feature map extraction network. The originality of the method proposed in this paper is as follows. First, we use a new feature map extraction network that is about 27% efficient than existing techniques in terms of memory. The proposed network does not reduce the size to the middle of the deep learning network, so important information required for 3D point cloud reconstruction is not lost. We solved the memory increase problem caused by the non-reduced image size by reducing the number of channels and by efficiently configuring the deep learning network to be shallow. Second, by preserving the high-resolution features of the 2D image, the accuracy can be further improved than that of the conventional technique. The feature map extracted from the non-reduced image contains more detailed information than the existing method, which can further improve the reconstruction accuracy of the 3D point cloud. Third, we use a divergence loss that does not require shooting information. The fact that not only the 2D image but also the shooting angle is required for learning, the dataset must contain detailed information and it is a disadvantage that makes it difficult to construct the dataset. In this paper, the accuracy of the reconstruction of the 3D point cloud can be increased by increasing the diversity of information through randomness without additional shooting information. In order to objectively evaluate the performance of the proposed method, using the ShapeNet dataset and using the same method as in the comparative papers, the CD value of the method proposed in this paper is 5.87, the EMD value is 5.81, and the FLOPs value is 2.9G. It was calculated. On the other hand, the lower the CD and EMD values, the better the accuracy of the reconstructed 3D point cloud approaches the original. In addition, the lower the number of FLOPs, the less memory is required for the deep learning network. Therefore, the CD, EMD, and FLOPs performance evaluation results of the proposed method showed about 27% improvement in memory and 6.3% in terms of accuracy compared to the methods in other papers, demonstrating objective performance.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.409-412
• /
• 2022

본 논문에서는 대입수시전형에서 생활기록부 입력은 대한민국 입시를 결정하는 매우 중요한 평가자료이다. 30명의 교사를 대상으로 실시한 설문조사를 통해서 교사들이 생활기록부를 작성하는데 가장 많이 참고하는 자료로는 수행평가 결과물, 발표내용, 보고서, 감상문 등의 학습 결과물(90%), 학생들이 작성한 자기평가서(73.3%), 관찰 평가지(50%)로 나타났으며, 80%(24명)의 교사들이 생활기록부를 작성하는데 고충을 겪고 있음을 확인할 수 있었다. 교사들이 느끼는 고충의 원인으로는 학생들의 개인별 특성 파악이 어려워 차별성있게 작성하는 것(76.7%)을 가장 힘들어 하였고, 작성해야 할 많은 수의 학생(60%), 문구를 만드는데 대해 부담(86.7%)을 느끼는 것으로 나타났다. 이 과정에서 교사의 전문성 뿐만 아니라 기계적이고 반복적인 작업도 많이 요구되고 있기 때문에, 생활기록부를 작성하는데에 도움을 줄 수 있는 프로그램 개발이 필요하다고 고안을 내었다. 교사들 역시 반복적이고 일률적인 생활기록부 작성에 도움을 줄 수 있는 프로그램이 있다면 유용하게 활용할 것이라는 응답이 90%였다. 따라서 본 연구에서 자연어 처리 인공지능 기술을 활용하여 교사들이 생활기록부를 작성하는데 있어 기계적이고 단순한 작업을 도와 주는 프로그램 개발에 대한 연구의 필요성을 제시하였다. 제안하는 프로그램은 학생들의 탐구보고서, 토론, 발표, 감상문 등의 생화기록부 작성 참고자료들을 텍스트로 변환하고 추상요약(Abstractive Summarization)을 통해 교사들이 효율적으로 작성하는데 활용될 수 있도록 설계하였다. 연구 결과 생활기록부 작성 참고자료를 텍스트로 변환하는 것과 추상요약을 할 수 있는 개방형 데이터셋까지는 확보하였다. 추상요약을 구현하는 방법에 대해서는 보다 심도 있는 추가연구가 필요하였다. 이를 통해 교사들이 교육 본질에 더욱 충실할 수 있는 환경을 마련하고, 내실 있는 생활기록부 작성이 공교육 신뢰 제고에 밑바탕이 되고자 한다.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.8
• /
• pp.157-164
• /
• 2024

In this paper, we propose a discussion that the feasibility of deploying a deep learning-based detector on the resource-limited board. Although many studies evaluate the detector on machines with high-performed GPUs, evaluation on the board with limited computation resources is still insufficient. Therefore, in this work, we implement the deep-learning detectors and deploy them on the compact board by parsing and optimizing a detector. To figure out the performance of deep learning based detectors on limited resources, we monitor the performance of several detectors with different H/W resource. On COCO detection datasets, we compare and analyze the evaluation results of detection model in On-Board and the detection model in On-GPU in terms of several metrics with mAP, power consumption, and execution speed (FPS). To demonstrate the effect of applying our detector for the military area, we evaluate them on our dataset consisting of thermal images considering the flight battle scenarios. As a results, we investigate the strength of deep learning-based on-board detector, and show that deep learning-based vision models can contribute in the flight battle scenarios.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2010.10a
• /
• pp.129-132
• /
• 2010

Users can obtain useful information from large of data because of development of internet. Since these data rise continuously like stream data, it is required to extract the appropriate information efficiently for the user's needs. In the traditional scheme, they are acquired by processing the user queries after the occurred data are stored at a database. However, it is inefficient to process the user queries over the large volume of continuous data by using the traditional scheme. In this paper, we propose the Event-Profile Model to define the data occurrence on the internet as the events and the user's requirements as the profiles. We also propose and implement the filtering scheme to process the events and the profiles efficiently. We evaluate the performance of the proposed scheme and our experiments show that the new scheme outperforms the other on various dataset.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.29 no.4
• /
• pp.785-794
• /
• 2019

Recently Deep Learning technology, one of the fourth industrial revolution technologies, is used to identify the hidden meaning of network data that is difficult to detect in the security arena and to predict attacks. Property and quality analysis of data sources are required before selecting the deep learning algorithm to be used for intrusion detection. This is because it affects the detection method depending on the contamination of the data used for learning. Therefore, the characteristics of the data should be identified and the characteristics selected. In this paper, the characteristics of malware were analyzed using network data set and the effect of each feature on performance was analyzed when the deep learning model was applied. The traffic classification experiment was conducted on the comparison of characteristics according to network characteristics and 96.52% accuracy was classified based on the selected characteristics.

• Phonetics and Speech Sciences
• /
• v.12 no.4
• /
• pp.63-71
• /
• 2020

Speech recognition is one of the areas actively commercialized using deep learning and machine learning techniques. However, the majority of speech recognition systems on the market are developed on data with limited diversity of speakers and tend to perform well on typical adult speakers only. This is because most of the speech recognition models are generally learned using a speech database obtained from adult males and females. This tends to cause problems in recognizing the speech of the elderly, children and people with dialects well. To solve these problems, it may be necessary to retain big database or to collect a data for applying a speaker adaptation. However, this paper proposes that a new end-to-end speech recognition method consists of an acoustic augmented recurrent encoder and a transformer decoder with linguistic prediction. The proposed method can bring about the reliable performance of acoustic and language models in limited data conditions. The proposed method was evaluated to recognize Korean elderly and children speech with limited amount of training data and showed the better performance compared of a conventional method.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_3
• /
• pp.1707-1720
• /
• 2023

With the increase in satellite time series data, the utility of remote sensing data is growing. In the analysis of time series data, the relative positional accuracy between images has a significant impact on the results, making image registration essential for correction. In recent years, research on image registration has been increasing by applying deep learning, which outperforms existing image registration algorithms. To train deep learning-based registration models, a large number of image pairs are required. Additionally, creating a correlation map between the data of existing deep learning models and applying additional computations to extract registration points is inefficient. To overcome these drawbacks, this study developed a data augmentation technique for training image registration models and applied it to OffsetNet, a registration model that predicts the offset amount itself, to perform image registration for KOMSAT-2, -3, and -3A. The results of the model training showed that OffsetNet accurately predicted the offset amount for the test data, enabling effective registration of the master and slave images.