• Journal of Intelligence and Information Systems
• /
• v.29 no.1
• /
• pp.203-214
• /
• 2023

The Korea Customs Service is efficiently handling business with an electronic customs system that can effectively handle one-stop business. This is the case and a more effective method is needed. Import and export require HS Code (Harmonized System Code) for classification and tax rate application for all goods, and item classification that classifies the HS Code is a highly difficult task that requires specialized knowledge and experience and is an important part of customs clearance procedures. Therefore, this study uses various types of data information such as product name, product description, and product image in the item classification request form to learn and develop a deep learning model to reflect information well based on Multimodal representation learning. It is expected to reduce the burden of customs duties by classifying and recommending HS Codes and help with customs procedures by promptly classifying items.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.8
• /
• pp.999-1005
• /
• 2020

While search portals' 'Portal News' account for the largest portion of aggregated news outlet, its neutrality as an outlet is questionable. This is because news aggregation may lead to prejudiced information consumption by recommending biased news articles. In this paper we introduce a new method of measuring political bias of news articles by using deep learning. It can provide its readers with insights on critical thinking. For this method, we build the dataset for deep learning by analyzing articles' bias from keywords, sourced from the National Assembly proceedings, and assigning bias to said keywords. Based on these data, news article bias is calculated by applying deep learning with a combination of Convolution Neural Network and Recurrent Neural Network. Using this method, 95.6% of sentences are correctly distinguished as either conservative or progressive-biased; on the entire article, the accuracy is 46.0%. This enables analyzing any articles' bias between conservative and progressive unlike previous methods that were limited on article subjects.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2018.06a
• /
• pp.190-192
• /
• 2018

최근 딥 러닝 기술의 발전이 가속화됨에 따라, 기존의 알고리즘과 융합하여 뛰어난 성능 향상을 보이는 연구가 급격히 증가하고 있다. 본 논문에서는 딥 러닝을 이용하여 HEVC 로 압축된 동영상의 일부 프레임의 삭제여부를 검출하는 알고리즘을 제안한다. 영상의 삭제 정보가 포함되어 있는 HEVC 의 부호화 파라미터를 추출하여 간단한 전 처리 과정을 통해 데이터의 크기를 효과적으로 압축한 뒤, 동영상의 시간적 특성을 고려할 수 있도록 CNN 네트워크를 구성한다. 실험 결과, 효과적으로 다양한 압축 환경에 강인한 영상 삭제 검출 성능을 보이는 것을 확인하였다.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.6
• /
• pp.1099-1110
• /
• 2023

The recognition system of autonomous driving and robot navigation performs vision work such as object recognition, tracking, and lane detection after multi-sensor fusion to improve performance. Currently, research on a deep learning model based on the fusion of a camera and a lidar sensor is being actively conducted. However, deep learning models are vulnerable to adversarial attacks through modulation of input data. Attacks on the existing multi-sensor-based autonomous driving recognition system are focused on inducing obstacle detection by lowering the confidence score of the object recognition model.However, there is a limitation that an attack is possible only in the target model. In the case of attacks on the sensor fusion stage, errors in vision work after fusion can be cascaded, and this risk needs to be considered. In addition, an attack on LIDAR's point cloud data, which is difficult to judge visually, makes it difficult to determine whether it is an attack. In this study, image scaling-based camera-lidar We propose an attack method that reduces the accuracy of LCCNet, a fusion model (camera-LiDAR calibration model). The proposed method is to perform a scaling attack on the point of the input lidar. As a result of conducting an attack performance experiment by size with a scaling algorithm, an average of more than 77% of fusion errors were caused.

• Journal of the Korean Society of Industry Convergence
• /
• v.23 no.2_2
• /
• pp.281-288
• /
• 2020

In this paper, we propose a control method to solve the surface hardness non-uniformity due to flow non-uniformity occurring in the heat treatment process of marine CAM. In the water cooling method including the decarbonization method, an automation device for deformation control has been developed and applied. LSTM was used to estimate the water cooling conditions, and the proposed method was found to be meaningful by improving the prototype results.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2020.05a
• /
• pp.500-503
• /
• 2020

특허를 통해 기술의 권리를 정의하고 보호하는 일이 매우 중요해짐에 따라 특허 문서를 분석하는 연구 또한 중요해지고 있다. 특히 특허의 청구항을 종속항과 독립항을 구분하고, 관련된 인용을 찾아내는 일은 관련 특허들을 분석하는데 매우 중요하다. 본 연구는 최근 텍스트 분석 분야에 획기적 성능 개선을 이끈 BERT(Bidirectional Encoder Representations From Transformers) 언어 모델을 사용하고 Neural Network 의 파인 튜닝 과정을 통해 청구항의 독립과 종속을 구분하였고, 인용하는 항의 번호와 인용 문구로 이루어진 인용 패턴을 통해 종속항의 인용 항을 찾아내었다. 이 방법을 2003 년 이후의 xml 형식의 미국 특허 데이터에 사용한 결과, 정확도 99% 의 성능을 확보하였다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.35-36
• /
• 2024

전신 마취 수술 중 저혈압의 발생은 다양한 합병증을 유발하며 이를 사전에 예측하여 대응하는 것은 매우 중요한 일이다. 따라서 본 연구에서는 SHAP 모델을 통해 변수 선택을 진행하고, Transformer 모델을 이용해 저혈압 발생 여부를 예측함으로써 임상적 의사결정을 지원한다. 또한 기존 연구들과는 달리, 수술실에서 수집되는 데이터를 기반으로 하여 높은 범용성을 가진다. 비침습적 혈압 예측에서 RMSE 9.46, MAPE 4.4%를 달성하였고, 저혈압 여부를 예측에서는 저혈압 기준 F1-Score 0.75로 우수한 결과를 얻었다.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.42 no.2
• /
• pp.263-272
• /
• 2022

As a result of climate change, the heatwave and urban heat island phenomena have become more common, and the frequency of heatwaves is expected to increase by two to six times by the year 2050. In particular, the heat sensation index felt by workers at construction sites during a heatwave is very high, and the sensation index becomes even higher if the urban heat island phenomenon is considered. The construction site environment and the situations of construction workers vulnerable to heat are not improving, and it is now imperative to respond effectively to reduce such damage. In this study, satellite imagery, land surface temperatures (LST), and long short-term memory (LSTM) were applied to analyze areas above 33 ℃, with the most vulnerable areas with increased synergistic damage from heat waves and the urban heat island phenomena then predicted. It is expected that the prediction results will ensure the safety of construction workers and will serve as the basis for a construction site early-warning system.

• The Journal of Bigdata
• /
• v.6 no.2
• /
• pp.169-179
• /
• 2021

Consumers' food consumption behavior is likely to be affected not only by structured data such as consumer panel data but also by unstructured data such as mass media and social media. In this study, a deep learning-based consumption prediction model is generated and verified for the fusion data set linking structured data and unstructured data related to food consumption. The results of the study showed that model accuracy was improved when combining structured data and unstructured data. In addition, unstructured data were found to improve model predictability. As a result of using the SHAP technique to identify the importance of variables, it was found that variables related to blog and video data were on the top list and had a positive correlation with the amount of paprika purchased. In addition, according to the experimental results, it was confirmed that the machine learning model showed higher accuracy than the deep learning model and could be an efficient alternative to the existing time series analysis modeling.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.6
• /
• pp.173-178
• /
• 2022

Recently, a recommendation system using deep learning in social network services has been actively studied. However, in the case of a recommendation system using deep learning, the cold start problem and the increased learning time due to the complex computation exist as the disadvantage. In this paper, the user-tailored exercise routine recommendation algorithm is proposed using the user's metadata. Metadata (the user's height, weight, sex, etc.) set as the input of the model is applied to the designed model in the proposed algorithms. The exercise recommendation system model proposed in this paper is designed based on the neural collaborative filtering (NCF) algorithm using multi-layer perceptron and matrix factorization algorithm. The learning proceeds with proposed model by receiving user metadata and exercise information. The model where learning is completed provides recommendation score to the user when a specific exercise is set as the input of the model. As a result of the experiment, the proposed exercise recommendation system model showed 10% improvement in recommended performance and 50% reduction in learning time compared to the existing NCF model.