• 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.

• Journal of Information Processing Systems
• /
• v.17 no.2
• /
• pp.242-252
• /
• 2021

Following the intuition that the local information in time instances is hardly incorporated into the posterior sequence in long short-term memory (LSTM), this paper proposes an attention augmented mechanism for fault diagnosis of the complex chemical process data. Unlike conventional fault diagnosis and classification methods, an attention mechanism layer architecture is introduced to detect and focus on local temporal information. The augmented deep network results preserve each local instance's importance and contribution and allow the interpretable feature representation and classification simultaneously. The comprehensive comparative analyses demonstrate that the developed model has a high-quality fault classification rate of 95.49%, on average. The results are comparable to those obtained using various other techniques for the Tennessee Eastman benchmark process.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.201-203
• /
• 2022

Over the past five years, the number of patients treated for diabetes has increased by 27.7% to 3.22 million, and since blood sugar is still checked through finger blood collection, continuous blood glucose measurement and blood sugar peak confirmation are difficult and painful. To solve this problem, based on blood sugar data measured for 14 days, three months of blood sugar prediction data are provided to diabetics using artificial intelligence technology.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.28-31
• /
• 2021

전세계적으로 우울증은 정신 건강 질환으로써 문제가 되고 있으며, 이를 해결하기 위해 일상생활에서의 우울증 탐지에 대한 연구가 진행되고 있다. 따라서 본 논문에서는 일상생활에 밀접하게 연관되어 있는 AI 스피커를 사용한 어텐션 메커니즘(Attention Mechanism) 기반 멀티모달 우울증 감지 시스템을 제안한다. 제안된 방법은 AI 스피커로부터 수집할 수 있는 음성 및 텍스트 데이터를 수집하고 CNN(Convolutional Neural Network)과 BiLSTM(Bidirectional Long Short-Term Memory Network)를 통해 각 데이터에서의 학습을 진행한다. 학습과정에서 Self-Attention 을 적용하여 특징 벡터에 추가적인 가중치를 부여하는 어텐션 메커니즘을 사용한다. 최종적으로 음성 및 텍스트 데이터에서 어텐션 가중치가 추가된 특징들을 합하여 SoftMax 를 통해 우울증 점수를 예측한다.

• Journal of Intelligence and Information Systems
• /
• v.22 no.2
• /
• pp.127-142
• /
• 2016

Deep learning model is a kind of neural networks that allows multiple hidden layers. There are various deep learning architectures such as convolutional neural networks, deep belief networks and recurrent neural networks. Those have been applied to fields like computer vision, automatic speech recognition, natural language processing, audio recognition and bioinformatics where they have been shown to produce state-of-the-art results on various tasks. Among those architectures, convolutional neural networks and recurrent neural networks are classified as the supervised learning model. And in recent years, those supervised learning models have gained more popularity than unsupervised learning models such as deep belief networks, because supervised learning models have shown fashionable applications in such fields mentioned above. Deep learning models can be trained with backpropagation algorithm. Backpropagation is an abbreviation for "backward propagation of errors" and a common method of training artificial neural networks used in conjunction with an optimization method such as gradient descent. The method calculates the gradient of an error function with respect to all the weights in the network. The gradient is fed to the optimization method which in turn uses it to update the weights, in an attempt to minimize the error function. Convolutional neural networks use a special architecture which is particularly well-adapted to classify images. Using this architecture makes convolutional networks fast to train. This, in turn, helps us train deep, muti-layer networks, which are very good at classifying images. These days, deep convolutional networks are used in most neural networks for image recognition. Convolutional neural networks use three basic ideas: local receptive fields, shared weights, and pooling. By local receptive fields, we mean that each neuron in the first(or any) hidden layer will be connected to a small region of the input(or previous layer's) neurons. Shared weights mean that we're going to use the same weights and bias for each of the local receptive field. This means that all the neurons in the hidden layer detect exactly the same feature, just at different locations in the input image. In addition to the convolutional layers just described, convolutional neural networks also contain pooling layers. Pooling layers are usually used immediately after convolutional layers. What the pooling layers do is to simplify the information in the output from the convolutional layer. Recent convolutional network architectures have 10 to 20 hidden layers and billions of connections between units. Training deep learning networks has taken weeks several years ago, but thanks to progress in GPU and algorithm enhancement, training time has reduced to several hours. Neural networks with time-varying behavior are known as recurrent neural networks or RNNs. A recurrent neural network is a class of artificial neural network where connections between units form a directed cycle. This creates an internal state of the network which allows it to exhibit dynamic temporal behavior. Unlike feedforward neural networks, RNNs can use their internal memory to process arbitrary sequences of inputs. Early RNN models turned out to be very difficult to train, harder even than deep feedforward networks. The reason is the unstable gradient problem such as vanishing gradient and exploding gradient. The gradient can get smaller and smaller as it is propagated back through layers. This makes learning in early layers extremely slow. The problem actually gets worse in RNNs, since gradients aren't just propagated backward through layers, they're propagated backward through time. If the network runs for a long time, that can make the gradient extremely unstable and hard to learn from. It has been possible to incorporate an idea known as long short-term memory units (LSTMs) into RNNs. LSTMs make it much easier to get good results when training RNNs, and many recent papers make use of LSTMs or related ideas.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1328-1336
• /
• 2019

To maintain the safe and optimal performance of batteries, accurate estimation of state of charge (SOC) is critical. In this paper, Long short-term memory network (LSTM) based on the artificial intelligence algorithm is applied to address the problem of the conventional coulomb-counting method. Different discharge cycles are concatenated to form the dataset for training and verification. In oder to improve the quality of input data for learning, preprocessing was performed. In addition, we compared learning ability and SOC estimation performance according to the structure of LSTM model and hyperparameter setup. The trained model was verified with a UDDS profile and achieved estimated accuracy of RMSE 0.82% and MAX 2.54%.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.7
• /
• pp.3599-3619
• /
• 2019

In human activity recognition system both static and motion information play crucial role for efficient and competitive results. Most of the existing methods are insufficient to extract video features and unable to investigate the level of contribution of both (Static and Motion) components. Our work highlights this problem and proposes Static-Motion fused features descriptor (SMFD), which intelligently leverages both static and motion features in the form of descriptor. First, static features are learned by two-stream 3D convolutional neural network. Second, trajectories are extracted by tracking key points and only those trajectories have been selected which are located in central region of the original video frame in order to to reduce irrelevant background trajectories as well computational complexity. Then, shape and motion descriptors are obtained along with key points by using SIFT flow. Next, cholesky transformation is introduced to fuse static and motion feature vectors to guarantee the equal contribution of all descriptors. Finally, Long Short-Term Memory (LSTM) network is utilized to discover long-term temporal dependencies and final prediction. To confirm the effectiveness of the proposed approach, extensive experiments have been conducted on three well-known datasets i.e. UCF101, HMDB51 and YouTube. Findings shows that the resulting recognition system is on par with state-of-the-art methods.

• Journal of Digital Contents Society
• /
• v.17 no.3
• /
• pp.157-164
• /
• 2016

Solar power generation influenced by the weather. Using the weather forecast information, it is possible to predict the short-term solar power generation in the future. However, in limited circumstances such as islands or mountains, it can not be use weather forecast information by the disconnection of the network, it is impossible to use solar power generation prediction model using weather forecast. Therefore, in this paper, we propose a system that can predict the short-term solar power generation by using the information that can be collected by the system itself. We developed a short-term prediction model using the prior information of temperature and power generation amount to improve the accuracy of the prediction. We showed the usefulness of proposed prediction model by applying to actual solar power generation data.

• Journal of Biomedical Engineering Research
• /
• v.40 no.6
• /
• pp.242-249
• /
• 2019

Forearm electromyography (EMG) generated by wrist movements has been widely used to develop an electrical prosthetic hand, but EMG generated by finger movements has been rarely used even though 20% of amputees lose fingers. The goal of this study is to improve the classification performance of different finger movements using a deep learning algorithm, and thereby contributing to the development of a high-performance finger-based prosthetic hand. Ten participants took part in this study, and they performed seven different finger movements forty times each (thumb, index, middle, ring, little, fist and rest) during which EMG was measured from the back of the right hand using four bipolar electrodes. We extracted mean absolute value (MAV), root mean square (RMS), and mean (MEAN) from the measured EMGs for each trial as features, and a 5x5-fold cross-validation was performed to estimate the classification performance of seven different finger movements. A long short-term memory (LSTM) model was used as a classifier, and linear discriminant analysis (LDA) that is a widely used classifier in previous studies was also used for comparison. The best performance of the LSTM model (sensitivity: 91.46 ± 6.72%; specificity: 91.27 ± 4.18%; accuracy: 91.26 ± 4.09%) significantly outperformed that of LDA (sensitivity: 84.55 ± 9.61%; specificity: 84.02 ± 6.00%; accuracy: 84.00 ± 5.87%). Our result demonstrates the feasibility of a deep learning algorithm (LSTM) to improve the performance of classifying different finger movements using EMG.

• Journal of the Korean Data and Information Science Society
• /
• v.24 no.3
• /
• pp.585-592
• /
• 2013

As the needs for alternative energy and renewable energy increase, there has been a lot of investment in developing wind energy, which does not cause air pollution nor the greenhouse gas effect. Wind energy is an environment friendly energy that is unlimited in its resources and is possible to be produced wherever the wind blows. However, since wind energy heavily relies on wind that has unreliable characteristics, it may be difficult to have efficient energy transmissions. For this reason, an important factor in wind energy forecasting is the estimation of available wind power. In this study, Gunsan wind farm data was used to compare ARMA model to neural network model to analyze for more accurate prediction of wind power generation. As a result, the neural network model was better than the ARMA model in the accuracy of the wind power predictions.