• Annual Conference on Human and Language Technology
• /
• 2015.10a
• /
• pp.194-197
• /
• 2015

Long Short-term Memory Network(LSTM) 기반 Recurrent Neural Network(RNN)는 순차 데이터를 모델링 할 수 있는 딥 러닝 모델이다. 기존 RNN의 그래디언트 소멸 문제(vanishing gradient problem)를 해결한 LSTM RNN은 멀리 떨어져 있는 이전의 입력 정보를 볼 수 있다는 장점이 있어 음성 인식 및 필기체 인식 등의 분야에서 좋은 성능을 보이고 있다. 또한 LSTM RNN 모델에 의존성(전이 확률)을 추가한 LSTM CRF모델이 자연어처리의 한 분야인 개체명 인식에서 우수한 성능을 보이고 있다. 본 논문에서는 한국어 문장의 지배소가 문장 후위에 나타나는 점에 착안하여 Backward 방식의 LSTM CRF 모델을 제안하고 이를 한국어 의미역 결정에 적용하여 기존 연구보다 더 높은 성능을 얻을 수 있음을 보인다.

• Annual Conference on Human and Language Technology
• /
• 2015.10a
• /
• pp.224-227
• /
• 2015

상호참조해결은 특정 개체에 대해 다르게 표현한 단어들을 서로 연관지어 주며, 이러한 개체에 대해 표현한 단어들을 멘션(mention)이라 하며, 이런 멘션을 찾아내는 것을 멘션탐지(mention detection)라 한다. 멘션은 명사나 명사구를 기반으로 정의되며, 명사구의 경우에는 수식어를 포함하기 때문에 멘션탐지를 순차 데이터 문제(sequence labeling problem)로 정의할 수 있다. 순차 데이터 문제에는 Recurrent Neural Network(RNN) 종류의 모델을 적용할 수 있으며, 모델들은 Long Short-Term Memory(LSTM) RNN, LSTM Recurrent CRF(LSTM-CRF), Bidirectional LSTM-CRF(Bi-LSTM-CRF) 등이 있다. LSTM-RNN은 기존 RNN의 그레디언트 소멸 문제(vanishing gradient problem)를 해결하였으며, LSTM-CRF는 출력 결과에 의존성을 부여하여 순차 데이터 문제에 더욱 최적화 하였다. Bi-LSTM-CRF는 과거입력자질과 미래입력자질을 함께 학습하는 방법으로 최근에 가장 좋은 성능을 보이고 있다. 이에 따라, 본 논문에서는 멘션탐지에 Bi-LSTM-CRF를 적용할 것을 제안하며, 각 딥 러닝 모델들에 대한 비교실험을 보인다.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.52 no.2
• /
• pp.102-106
• /
• 2003

This paper describes a new pattern classifier neural network to extract the feature from a letter. The proposed pattern classifier is based on relative distance, which is measure between an input datum and the center of cluster group. So, the proposed classifier neural network is called relative neural network(RNN). According to definitions of the distance and the learning rule, the structure of RNN is designed and the pseudo code of the algorithm is described. In feature extraction of letter, RNN, in spite of deletion of learning rate, resulted in the identical performance with those of winner-take-all(WTA), and self-organizing-map(SOM) neural network. Thus, it is shown that RNN is suitable to extract the feature of a letter.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.660-661
• /
• 2020

본 논문에서는 기존 인공 신경망 기반 시계열 학습 기법인 Recurrent Neural Network (RNN)의 많은 연산량 및 고 사양 시스템 요구를 개선하기 위해 랜덤 포레스트 (Random Forest)기반의 새로운 시계열 학습 기법을 제안한다. 기존의 RNN 기반 방법들은 복잡한 연산을 통해 높은 성능을 달성하는 데 집중하고 있다. 이러한 방법들은 학습에 많은 파라미터가 필요할 뿐만 아니라 대규모의 연산을 요구하므로 실시간 시스템에 적용하는데 어려움이 있다. 따라서 본 논문에서는, 효율적이면서 빠르게 동작할 수 있는 시계열 다층 랜덤 포레스트(Time series Multilayered Random Forest)를 제안하고 산불 조기 탐지에 적용해 기존 RNN 계열의 방법들과 성능을 비교하였다. 다양한 산불화재 실험데이터에 알고리즘을 적용해본 결과 GPU 상에서 방대한 연산을 수행하는 RNN 기반 방법들과 비교해 성능적인 한계가 존재했지만 CPU 에서도 빠르게 동작 가능하므로 성능의 개선을 통해 다양한 임베디드 시스템에 적용 가능하다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.473-474
• /
• 2023

본 논문에서는 딥러닝 기법 중 하나인 순환신경망(RNN)을 활용하여 자연어를 처리할 수 있는 모델 개발에 대하여 연구를 진행하였다. 다양한 주제에 대한 사용자들의 의견을 확보할 수 있는 유튜브 플랫픔을 활용하여 데이터를 확보하였으며, 감성 분류를 진행하는 만큼 학습 데이터셋으로는 네이버 영화 리뷰 데이터셋을 활용하였다. 사용자는 직접 데이터 파일을 삽입하거나 혹은 유튜브 댓글과 같이 데이터를 외부에서 확보하여 감성을 분석할 수 있으며, 자연어 속 등장하는 단어의 빈도수를 종합하여 해당 데이터들 속 키워드는 무엇인지를 분석할 수 있도록 하였다. 나아가 종합 데이터 분석 관리 플랫폼을 제작하기 위하여 해당 데이터를 데이터베이스에 저장하고GUI 프로그램을 통하여 접근 및 관리가 가능하도록 하였다.

• Journal of Digital Contents Society
• /
• v.18 no.1
• /
• pp.93-99
• /
• 2017

A hybrid approach using Long Short Term Memory (LSTM) Recurrent Neural Network (RNN) has showed great improvement in speech recognition accuracy. For training acoustic model based on hybrid approach, it requires forced alignment of HMM state sequence from Gaussian Mixture Model (GMM)-Hidden Markov Model (HMM). However, high computation time for training GMM-HMM is required. This paper proposes an end-to-end approach for LSTM RNN-based Korean speech recognition to improve learning speed. A Connectionist Temporal Classification (CTC) algorithm is proposed to implement this approach. The proposed method showed almost equal performance in recognition rate, while the learning speed is 1.27 times faster.

• Journal of the Society of Disaster Information
• /
• v.3 no.1
• /
• pp.37-53
• /
• 2007

Various difficult problems occur due to insufficient bearing capacity or excessive settlements when constructing roads or large complexes. Accurate predictions on the final settlement and consolidation time can help in choosing the ground improvement method and thus enables to save time and expense of the whole project. Asaoka's method is probably the most frequently used for settlement prediction which are based on Terzaghi's one dimensional consolidation theory. Empirical formulae such as Hyperbolic method and Hoshino's method are also often used. However, it is known that the settlement predicted by these methods do not match with the actual settlements. Furthermore these methods cannot be used at design stage when there is no measured data. To find an elaborate method in predicting settlement in embankments using various test results and actual settlement data from domestic sites, Back-Propagation Neural Network(BPNN) and Recurrent Neural Network(RNN) were employed and the most suitable model structures were obtained. Predicted settlement values by the developed models were compared with the measured values as well as numerical analysis results. Analysis of the results showed that RNN yielded more compatible predictions with actual data than BPNN and predictions using cone penetration resistance were closer to actual data than predictions using SPT results. Also, it was found that the developed method were very competitive with the numerical analysis considering the number of input data, complexity and effort in modelling. It is believed that RNN using cone penetration test results can make a highly efficient tool in predicting settlements if enough field data can be obtained.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.7
• /
• pp.259-266
• /
• 2018

In this paper, we propose a method to classify a document using a Recurrent Neural Network by extracting features considering word sense and contexts. Word2vec method is adopted to include the order and meaning of the words expressing the word in the document as a vector. Doc2vec is applied for considering the context to extract the feature of the document. RNN classifier, which includes the output of the previous node as the input of the next node, is used as the document classification method. RNN classifier presents good performance for document classification because it is suitable for sequence data among neural network classifiers. We applied GRU (Gated Recurrent Unit) model which solves the vanishing gradient problem of RNN. It also reduces computation speed. We used one Hangul document set and two English document sets for the experiments and GRU based document classifier improves performance by about 3.5% compared to CNN based document classifier.

• Journal of the Korean Geotechnical Society
• /
• v.18 no.5
• /
• pp.123-132
• /
• 2002

The purpose of this study is to predict the shear behavior of the weathered mudstone soil using dynamic neural network which mimics the biological system of human brain. SNN and RNN, which are kinds of the dynamic neural network realizing continuously a pattern recognition as time goes by, are used to predict a nonlinear behavior of soil. After analysis, parameters which have an effect on learning and predicting of neural network, the teaming rate, momentum constant and the optimum neural network model are decided to be 0.5, 0.7, 8$\times$18$\times$2 in SU model and 0.3, 0.9, 8$\times$24$\times$2 in R model. The results of appling both networks showed that both networks predicted the shear behavior of soil in normally consolidated state well, but RNN model which is effective fir input data of irregular patterns predicted more efficiently than SNN model in case of the prediction in overconsolidated state.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.4
• /
• pp.494-500
• /
• 2020

This paper proposes a new distance estimation technique for indoor localization in ultra wideband (UWB) systems. The proposed technique is based on recurrent neural network (RNN), one of the deep learning methods. The RNN is known to be useful to deal with time series data, and since UWB signals can be seen as a time series data, RNN is employed in this paper. Specifically, the transmitted UWB signal passes through IEEE802.15.4a indoor channel model, and from the received signal, the RNN regressor is trained to estimate the distance from the transmitter to the receiver. To verify the performance of the trained RNN regressor, new received UWB signals are used and the conventional threshold based technique is also compared. For the performance measure, root mean square error (RMSE) is assessed. According to the computer simulation results, the proposed distance estimator is always much better than the conventional technique in all signal-to-noise ratios and distances between the transmitter and the receiver.