• 한국정보통신학회논문지
• /
• 제26권2호
• /
• pp.187-192
• /
• 2022

위드 코로나의 예정 발표일이 결정되었고, 위드 코로나에 가장 중요한 조건인 백신 접종을 아직 부작용 걱정 때문에 완료하지 않은 사람들이 있다. 또한 위드 코로나로 경제는 회복될 수 있지만 감염자 수는 급증할 수 있다. 본 논문은 위드 코로나에 앞서 코로나19에 대한 경각심을 깨우고자, 코로나19를 비선형 확률과정으로 예측한다. 여기서 딥러닝의 RNN중 양방향 LSTM인 BI-LSTM와 LSTM보다 gate수를 줄인 GRU를 사용하고 이것을 train set, test set, 손실함수, 잔차분석, 정규분포, 자기 상관을 통해서 비교 분석하여 어떠한 성능이 더 좋은지 비교하고 예측한다.

• 한국컴퓨터정보학회논문지
• /
• 제28권7호
• /
• pp.11-18
• /
• 2023

본 논문에서는 딥러닝 시계열 예측 모형을 평가한다. 최근 연구에 따르면 이 모형은 ARIMA와 같은 기존 예측 모형보다 성능이 우수하다고 결론짓는다. 그 중 히든 레이어에 이전 정보를 저장하는 순환 신경망이 이를 위한 예측 모형 중 하나이다. 네트워크의 그래디언트 소실 문제를 해결하기 위해 LSTM은 데이터 흐름의 반대 방향으로 숨겨진 레이어가 추가되는 BI-LSTM과 함께 순환 신경망 내부의 작은 메모리로 사용된다. 본 논문은 서울의 2018년 1월 1일부터 2022년도 1월 1일까지의 NO2 자료에 대해 Informer의 성능을 LSTM, BI-LSTM, Transformer와 비교하였다. 이에 실제 값과 예측값 사이의 평균 제곱근 오차와 평균 절대 오차를 구하였다. 그 결과 Test 데이터(2021.09.01.~2022.01.01.)에 대해 Informer는 다른 방법에 비해 가장 높은 예측 정확도 (가장 낮은 예측 오차: 평균 제곱근 오차: 0.0167, 평균 절대 오차: 0.0138)를 보여 타 방법에 비해 그 우수성을 입증하였다. Informer는 당초 취지와 부합되게 다른 방법들이 갖고 있는 장기 시계열 예측에 있어서의 문제점을 개선하는 결과를 나타내고 있다.

• ETRI Journal
• /
• 제44권3호
• /
• pp.438-449
• /
• 2022

This study proposes a dependency grammar-based self-attention multilayered bidirectional long short-term memory (DG-M-Bi-LSTM) model for subject-predicate-object (SPO) tuple recognition from natural language (NL) sentences. To add recent knowledge to the knowledge base autonomously, it is essential to extract knowledge from numerous NL data. Therefore, this study proposes a high-accuracy SPO tuple recognition model that requires a small amount of learning data to extract knowledge from NL sentences. The accuracy of SPO tuple recognition using DG-M-Bi-LSTM is compared with that using NL-based self-attention multilayered bidirectional LSTM, DG-based bidirectional encoder representations from transformers (BERT), and NL-based BERT to evaluate its effectiveness. The DG-M-Bi-LSTM model achieves the best results in terms of recognition accuracy for extracting SPO tuples from NL sentences even if it has fewer deep neural network (DNN) parameters than BERT. In particular, its accuracy is better than that of BERT when the learning data are limited. Additionally, its pretrained DNN parameters can be applied to other domains because it learns the structural relations in NL sentences.

• 한국정보기술학회 영문논문지
• /
• 제10권2호
• /
• pp.167-175
• /
• 2020

In this paper, we propose a traffic accident detection on vehicle-mounted camera. In the proposed method, the minimum bounding box coordinates the central coordinates on the bird's eye view and motion vectors of each vehicle object, and ego-motions of the vehicle equipped with dash-cam are extracted from the dash-cam video. By using extracted 4 kinds features as the input of Bi-LSTM (bidirectional LSTM), the accident probability (score) is predicted. To investigate the effect of each input feature on the probability of an accident, we analyze the performance of the detection the case of using a single feature input and the case of using a combination of features as input, respectively. And in these two cases, different detection models are defined and used. Bi-LSTM is used as a cascade, especially when a combination of the features is used as input. The proposed method shows 76.1% precision and 75.6% recall, which is superior to our previous work.

• International Journal of Internet, Broadcasting and Communication
• /
• 제13권2호
• /
• pp.173-178
• /
• 2021

In this paper, we find a suitable methodology for Korean Sentiment Analysis through a comparative experiment in which methods of embedding and natural network models are learned at the highest accuracy and fastest speed. The embedding method compares word embeddeding and Word2Vec. The model compares and experiments representative neural network models CNN, RNN, LSTM, GRU, Bi-LSTM and Bi-GRU with IKEA review data. Experiments show that Word2Vec and BiGRU had the highest accuracy and second fastest speed with 94.23% accuracy and 42.30 seconds speed. Word2Vec and GRU were found to have the third highest accuracy and fastest speed with 92.53% accuracy and 26.75 seconds speed.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.676-677
• /
• 2022

본 논문에서는 패션 의류 이미지의 빠르고 정확한 분류를 달성하기 위해 최적화된 동적 붕괴 학습률과 개선된 모델 구조를 가진 딥 러닝 모델을 기반으로 하는 새로운 방법을 제안한다. Fashion-MNIST 데이터 셋에서 제안된 모델을 사용하여 실험을 수행하고 CNN, LeNet, LSTM 및 BiLSTM의 방법과 비교한다.

• International Journal of Contents
• /
• 제17권4호
• /
• pp.1-15
• /
• 2021

Named entity recognition (NER) is a crucial task for NLP, which aims to extract information from texts. To build NER systems, deep learning (DL) models are learned with dictionary features by mapping each word in the dataset to dictionary features and generating a unique index. However, this technique might generate noisy labels, which pose significant challenges for the NER task. In this paper, we proposed DL-dictionary features, and evaluated them on two datasets, including the OntoNotes 5.0 dataset and our new infectious disease outbreak dataset named GFID. We used (1) a Bidirectional Long Short-Term Memory (BiLSTM) character and (2) pre-trained embedding to concatenate with (3) our proposed features, named the Convolutional Neural Network (CNN), BiLSTM, and self-attention dictionaries, respectively. The combined features (1-3) were fed through BiLSTM - Conditional Random Field (CRF) to predict named entity classes as outputs. We compared these outputs with other predictions of the BiLSTM character, pre-trained embedding, and dictionary features from previous research, which used the exact matching and partial matching dictionary technique. The findings showed that the model employing our dictionary features outperformed other models that used existing dictionary features. We also computed the F1 score with the GFID dataset to apply this technique to extract medical or healthcare information.

• 산업경영시스템학회지
• /
• 제46권4호
• /
• pp.173-180
• /
• 2023

Smart factory companies are installing various sensors in production facilities and collecting field data. However, there are relatively few companies that actively utilize collected data, academic research using field data is actively underway. This study seeks to develop a model that detects anomalies in the process by analyzing spindle power data from a company that processes shafts used in automobile throttle valves. Since the data collected during machining processing is time series data, the model was developed through unsupervised learning by applying the Holt Winters technique and various deep learning algorithms such as RNN, LSTM, GRU, BiRNN, BiLSTM, and BiGRU. To evaluate each model, the difference between predicted and actual values was compared using MSE and RMSE. The BiLSTM model showed the optimal results based on RMSE. In order to diagnose abnormalities in the developed model, the critical point was set using statistical techniques in consultation with experts in the field and verified. By collecting and preprocessing real-world data and developing a model, this study serves as a case study of utilizing time-series data in small and medium-sized enterprises.

• 반도체디스플레이기술학회지
• /
• 제22권3호
• /
• pp.41-45
• /
• 2023

Currently, the method of loading items in the warehouse, the worker directly decides the loading location, and the most used method is to load the product at the location closest to the entrance. This can be effective when there is no difference in the required amount for goods, but when there is a difference in the required amount for goods, it is inefficient because items with a small required amount are loaded near the entrance and occupy the corresponding space for a long time. Therefore, in order to minimize the release time of goods, it is essential to select an appropriate location when loading goods. In this study, a method for determining the loading location by predicting the required amount of goods was studied to select the optimal loading location. Deep learning based bidirectional long-term memory networks (Bi-LSTM) was used to predict the required amount for goods. This study compares and analyzes the release time of goods in the conventional method of loading close to the entrance and in the loading method using the required amount for goods using the Bi-LSTM model.

• International Journal of Contents
• /
• 제18권2호
• /
• pp.68-80
• /
• 2022

The health of the human heart is commonly measured using ECG (Electrocardiography) signals. To identify any anomaly in the human heart, the time-sequence of ECG signals is examined manually by a cardiologist or cardiac electrophysiologist. Lightweight anomaly detection on ECG signals in an embedded system is expected to be popular in the near future, because of the increasing number of heart disease symptoms. Some previous research uses deep learning networks such as LSTM and BiLSTM to detect anomaly signals without any handcrafted feature. Unfortunately, lightweight LSTMs show low precision and heavy LSTMs require heavy computing powers and volumes of labeled dataset for symptom classification. This paper proposes an ECG anomaly detection system based on two level BiLSTM for acceptable precision with lightweight networks, which is lightweight and usable at home. Also, this paper presents a new threshold technique which considers statistics of the current ECG pattern. This paper's proposed model with BiLSTM detects ECG signal anomaly in 0.467 ~ 1.0 F₁ score, compared to 0.426 ~ 0.978 F₁ score of the similar model with LSTM except one highly noisy dataset.