• ETRI Journal
• /
• v.43 no.6
• /
• pp.1038-1048
• /
• 2021

We propose an end-to-end neural coreference resolution for the Korean language that uses an attention mechanism to point to the same entity. Because Korean is a head-final language, we focused on a method that uses a pointer network based on the head. The key idea is to consider all nouns in the document as candidates based on the head-final characteristics of the Korean language and learn distributions over the referenced entity positions for each noun. Given the recent success of applications using bidirectional encoder representation from transformer (BERT) in natural language-processing tasks, we employed BERT in the proposed model to create word representations based on contextual information. The experimental results indicated that the proposed model achieved state-of-the-art performance in Korean language coreference resolution.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.3
• /
• pp.115-122
• /
• 2019

For Korean phoneme recognition, Hidden Markov-Gaussian Mixture model(HMM-GMM) or hybrid models which combine artificial neural network with HMM have been mainly used. However, current approach has limitations in that such models require force-aligned corpus training data that is manually annotated by experts. Recently, researchers used neural network based phoneme recognition model which combines recurrent neural network(RNN)-based structure with connectionist temporal classification(CTC) algorithm to overcome the problem of obtaining manually annotated training data. Yet, in terms of implementation, these RNN-based models have another difficulty in that the amount of data gets larger as the structure gets more sophisticated. This problem of large data size is particularly problematic in the Korean language, which lacks refined corpora. In this study, we introduce CTC algorithm that does not require force-alignment to create a Korean phoneme recognition model. Specifically, the phoneme recognition model is based on convolutional neural network(CNN) which requires relatively small amount of data and can be trained faster when compared to RNN based models. We present the results from two different experiments and a resulting best performing phoneme recognition model which distinguishes 49 Korean phonemes. The best performing phoneme recognition model combines CNN with 3hop Bidirectional LSTM with the final Phoneme Error Rate(PER) at 3.26. The PER is a considerable improvement compared to existing Korean phoneme recognition models that report PER ranging from 10 to 12.

• Phonetics and Speech Sciences
• /
• v.11 no.2
• /
• pp.57-64
• /
• 2019

In this paper, we construct a Korean text-to-speech system using the Merlin toolkit which is an open source system for speech synthesis. In the text-to-speech system, the HMM-based statistical parametric speech synthesis method is widely used, but it is known that the quality of synthesized speech is degraded due to limitations of the acoustic modeling scheme that includes context factors. In this paper, we propose an acoustic modeling architecture that uses deep neural network technique, which shows excellent performance in various fields. Fully connected deep feedforward neural network (DNN), recurrent neural network (RNN), gated recurrent unit (GRU), long short-term memory (LSTM), bidirectional LSTM (BLSTM) are included in the architecture. Experimental results have shown that the performance is improved by including sequence modeling in the architecture, and the architecture with LSTM or BLSTM shows the best performance. It has been also found that inclusion of delta and delta-delta components in the acoustic feature parameters is advantageous for performance improvement.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1295-1301
• /
• 2019

In this paper, we propose a novel method of denoising human motion using a bidirectional recurrent neural network (BRNN) with an attention mechanism. The corrupted motion captured from a single 3D depth sensor camera is automatically fixed in the well-established smooth motion manifold. Incorporating an attention mechanism into BRNN achieves better optimization results and higher accuracy than other deep learning frameworks because a higher weight value is selectively given to a more important input pose at a specific frame for encoding the input motion. Experimental results show that our approach effectively handles various types of motion and noise, and we believe that our method can sufficiently be used in motion capture applications as a post-processing step after capturing human motion.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.4
• /
• pp.427-432
• /
• 2019

In this paper, we propose a home monitoring system using sound event detection based on a bidirectional gated recurrent neural network for the hard-of-hearing. First, in the proposed system, packet loss concealment is used to recover a lost signal captured through wireless sensor networks, and reliable channels are selected using multi-channel cross correlation coefficient for effective sound event detection. The detected sound event is converted into the text and haptic signal through a harmonic/percussive sound source separation method to be provided to hearing impaired people. Experimental results show that the performance of the proposed sound event detection method is superior to the conventional methods and the sound can be expressed into detailed haptic signal using the source separation.

• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06b
• /
• pp.321-323
• /
• 2012

스마트폰 사용의 보편화와 센서기술의 발달로 이를 응용하는 다양한 연구가 진행되고 있다. 특히 가속도, GPS, 조도, 방향센서 등의 센서들이 스마트폰에 부착되어 출시되고 있어서, 이를 이용한 상황인지, 행동인식 등의 관련 연구들이 활발하다. 하지만 다양한 클래스를 분류하면서 높은 인식률을 유지하는 것은 어려운 문제이다. 본 논문에서는 인식률 향상을 위해 계층적 구조의 순환 신경망을 이용하여 제스처를 인식한다. 스마트폰의 가속도 센서를 이용하여 사용자의 제스처 데이터를 수집하고 BLSTM(Bidirectional Long Short-Term Memory) 구조의 순환신경망을 계층적으로 사용하여, 20가지 사용자의 제스처와 비제스처를 분류한다. 약 24,850개의 시퀀스 데이터를 사용하여 실험한 결과, 기존 BLSTM은 평균 89.17%의 인식률을 기록한 반면 계층적 BLSTM은 평균 91.11%의 인식률을 나타내었다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.4
• /
• pp.633-636
• /
• 2022

The reduction of energy consumption at the base station (BS) has become more important recently. In this paper, we consider the adaptive muting of the antennas based on the predicted future traffic load to reduce the energy consumption where the number of active antennas is adaptively adjusted according to the predicted future traffic load. Given that traffic load is sequential data, three different RNN structures, namely long-short term memory (LSTM), gated recurrent unit (GRU), and bidirectional LSTM (Bi-LSTM) are considered for the future traffic load prediction. Through the performance evaluation based on the actual traffic load collected from the Afghanistan telecom company, we confirm that the traffic load can be estimated accurately and the overall power consumption can also be reduced significantly using the antenna musing.

• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.159-162
• /
• 2017

Long Short-Term Memory Recurrent Neural Network(LSTM RNN)는 순차 데이터 모델링에 적합한 딥러닝 모델이다. Bidirectional LSTM RNN(BiLSTM RNN)은 RNN의 그래디언트 소멸 문제(vanishing gradient problem)를 해결한 LSTM RNN을 입력 데이터의 양 방향에 적용시킨 것으로 입력 열의 모든 정보를 볼 수 있는 장점이 있어 자연어처리를 비롯한 다양한 분야에서 많이 사용되고 있다. Highway Network는 비선형 변환을 거치지 않은 입력 정보를 히든레이어에서 직접 사용할 수 있게 LSTM 유닛에 게이트를 추가한 딥러닝 모델이다. 본 논문에서는 Highway Network를 한국어 의미역 결정에 적용하여 기존 연구 보다 더 높은 성능을 얻을 수 있음을 보인다.

• 한국어정보학회:학술대회논문집
• /
• 2017.10a
• /
• pp.159-162
• /
• 2017

Long Short-Term Memory Recurrent Neural Network(LSTM RNN)는 순차 데이터 모델링에 적합한 딥러닝 모델이다. Bidirectional LSTM RNN(BiLSTM RNN)은 RNN의 그래디언트 소멸 문제(vanishing gradient problem)를 해결한 LSTM RNN을 입력 데이터의 양 방향에 적용시킨 것으로 입력 열의 모든 정보를 볼 수 있는 장점이 있어 자연어처리를 비롯한 다양한 분야에서 많이 사용되고 있다. Highway Network는 비선형 변환을 거치지 않은 입력 정보를 히든레이어에서 직접 사용할 수 있게 LSTM 유닛에 게이트를 추가한 딥러닝 모델이다. 본 논문에서는 Highway Network를 한국어 의미역 결정에 적용하여 기존 연구 보다 더 높은 성능을 얻을 수 있음을 보인다.

• Geophysics and Geophysical Exploration
• /
• v.26 no.3
• /
• pp.126-137
• /
• 2023

In this study, a recurrent neural network (RNN) was employed as a methodological approach to classify dolphin click signals derived from ocean monitoring data. To improve the accuracy of click signal classification, the single time series data were transformed into fractional domains using fractional Fourier transform to expand its features. Transformed data were used as input for three RNN models: long short-term memory (LSTM), gated recurrent unit (GRU), and bidirectional LSTM (BiLSTM), which were compared to determine the optimal network for the classification of signals. Because the fractional Fourier transform displayed different characteristics depending on the chosen angle parameter, the optimal angle range for each RNN was first determined. To evaluate network performance, metrics such as accuracy, precision, recall, and F1-score were employed. Numerical experiments demonstrated that all three networks performed well, however, the BiLSTM network outperformed LSTM and GRU in terms of learning results. Furthermore, the BiLSTM network provided lower misclassification than the other networks and was deemed the most practically appliable to field data.