• 대한전기학회논문지:전력기술부문A
• /
• 제48권10호
• /
• pp.1231-1238
• /
• 1999

This paper presents an approach for the optimal heating load Identification using Diagonal Recurrent Neural Networks(DRNN). In this paper, the DRNN captures the dynamic nature of a system and since it is not fully connected, training is much faster than a fully connected recurrent neural network. The architecture of DRNN is a modified model of the fully connected recurrent neural network with one hidden layer. The hidden layer is comprised of self-recurrent neurons, each feeding its output only into itself. In this study, A dynamic backpropagation (DBP) with delta-bar-delta learning method is used to train an optimal heating load identifier. Delta-bar-delta learning method is an empirical method to adapt the learning rate gradually during the training period in order to improve accuracy in a short time. The simulation results based on experimental data show that the proposed model is superior to the other methods in most cases, in regard of not only learning speed but also identification accuracy.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 1999년도 추계종합학술대회 논문집
• /
• pp.1007-1010
• /
• 1999

This paper describes the recognition of totally unconstrained handwritten numerals using neural networks. Neural networks with multiple output nodes have been successfully used to classify complex handwritten numerals. The recognition system consists of the preprocessing stage to extract features using Kirsch mask and the classification stage to recognize the numerals using the fully-connected recurrent neural networks (RNN). Simulation results with the numeral database of Concordia university, Montreal, Canada, are presented. The recognition system proposed in this paper outperforms other recognition systems reported on the same database.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 학술대회 논문집 정보 및 제어부문
• /
• pp.349-351
• /
• 2004

The classical dynamic backpropagation learning algorithm has the problems of learning speed and the determine of learning parameter. The Extend Kalman Filter(EKF) is used effectively for a state estimation method for a non linear dynamic system. This paper presents a learning algorithm using Dual Extended Kalman Filter(DEKF) for Fully Recurrent Neural Network(FRNN). This DEKF learning algorithm gives the minimum variance estimate of the weights and the hidden outputs. The proposed DEKF learning algorithm is applied to the system identification of a nonlinear SISO system and compared with dynamic backpropagation learning algorithm.

• Journal of Electrical Engineering and information Science
• /
• 제1권1호
• /
• pp.108-117
• /
• 1996

In this paper, we propose a new type of recurrent neural network architecture in which each output unit is connected with itself and fully-connected with other output units and all hidden units. The proposed recurrent network differs from Jordan's and Elman's recurrent networks in view of functions and architectures because it was originally extended from the multilayer feedforward neural network for improving the discrimination and generalization power. We also prove the convergence property of learning algorithm of the proposed recurrent neural network and analyze the performance of the proposed recurrent neural network by performing recognition experiments with the totally unconstrained handwritten numeral database of Concordia University of Canada. Experimental results confirmed that the proposed recurrent neural network improves the discrimination and generalization power in recognizing spatial patterns.

• Journal of information and communication convergence engineering
• /
• 제8권3호
• /
• pp.267-272
• /
• 2010

For the last decade, recurrent neural networks (RNNs) have been commonly applied to communications channel equalization. The major problems of gradient-based learning techniques, employed to train recurrent neural networks are slow convergence rates and long training sequences. In high-speed communications system, short training symbols and fast convergence speed are essentially required. In this paper, the derivative-free Kalman filter, so called the unscented Kalman filter (UKF), for training a fully connected RNN is presented in a state-space formulation of the system. The main features of the proposed recurrent neural equalizer are fast convergence speed and good performance using relatively short training symbols without the derivative computation. Through experiments of nonlinear channel equalization, the performance of the RNN with a derivative-free Kalman filter is evaluated.

• 한국음향학회지
• /
• 제36권6호
• /
• pp.407-412
• /
• 2017

본 논문에서는 주목 메커니즘 기반의 심층 신경망을 사용한 음성 감정인식 방법을 제안한다. 제안하는 방식은 CNN(Convolution Neural Networks), GRU(Gated Recurrent Unit), DNN(Deep Neural Networks)의 결합으로 이루어진 심층 신경망 구조와 주목 메커니즘으로 구성된다. 음성의 스펙트로그램에는 감정에 따른 특징적인 패턴이 포함되어 있으므로 제안하는 방식에서는 일반적인 CNN에서 컨벌루션 필터를 tuned Gabor 필터로 사용하는 GCNN(Gabor CNN)을 사용하여 패턴을 효과적으로 모델링한다. 또한 CNN과 FC(Fully-Connected)레이어 기반의 주목 메커니즘을 적용하여 추출된 특징의 맥락 정보를 고려한 주목 가중치를 구해 감정인식에 사용한다. 본 논문에서 제안하는 방식의 검증을 위해 6가지 감정에 대해 인식 실험을 진행하였다. 실험 결과, 제안한 방식이 음성 감정인식에서 기존의 방식보다 더 높은 성능을 보였다.

• 한국지능시스템학회논문지
• /
• 제15권5호
• /
• pp.552-559
• /
• 2005

실시간 순환형 훈련 알고리즘(RTRL)과 같이 경사법에 의해 훈련되는 순환형 뉴럴 네트웍(RNN)은 수렴속도가 매우 느린 단점을 지니고 있다. 이 알고리즘은 또한 오차 역전달 처리과정에서 결코 쉽지 않은 미분 계산을 필요로 한다. 본 논문에서는 완전하게 결합된 RNN의 훈련을 위하여 소위 언센티드 칼만필터라고 불리우는 미분없는 칼만필터 훈련 알고리즘을 시스템의 상태공간 상에서 표현하였다. 미분없는 칼만필터 훈련 알고리즘은 순환형 뉴럴 네트웍 훈련시 미분 계산 없이 매우 빠른 수렴속도와 좋은 추정 성능을 보여준다. 비선형 채널 등화 실험을 통하여 미분 없는 칼만필터 훈련 알고리즘을 이용한 RNN의 성능이 향상되었음을 보였다.

• 한국컴퓨터그래픽스학회논문지
• /
• 제27권5호
• /
• pp.73-79
• /
• 2021

최근 컴퓨터 애니메이션 분야에서는 기존의 유한상태기계나 그래프 기반의 방식들에서 벗어나 딥러닝을 이용한 동작 생성 방식이 많이 연구되고있다. 동작 학습에 요구되는 네트워크의 표현력은 학습해야하는 동작의 단순한 길이보다는 그 안에 포함된 동작의 다양성에 더 큰 영향을 받는다. 본 연구는 이처럼 학습해야하는 동작의 종류가 다양한 경우에 효율적인 네트워크 구조를 찾는것을 목표로 한다. 기본적인 fully-connected 구조, 여러개의 fully-connected 레이어를 병렬적으로 사용하는 mixture of experts구조, seq2seq처리에 널리 사용되는 순환신경망(RNN), 그리고 최근 시퀀스 형태의 데이터 처리를 위해 자연어 처리 분야에서 사용되고있는 transformer구조의 네트워크들을 각각 학습하고 비교한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권9호
• /
• pp.3204-3220
• /
• 2021

We perform the task of video object segmentation by incorporating a conditional random field (CRF) and convolutional neural networks (CNNs). Most methods employ a CRF to refine a coarse output from fully convolutional networks. Others treat the inference process of the CRF as a recurrent neural network and then combine CNNs and the CRF into an end-to-end model for video object segmentation. In contrast to these methods, we propose a novel higher-order CRF model to solve the problem of video object segmentation. Specifically, we use CNNs to establish a higher-order dependence among pixels, and this dependence can provide critical global information for a segmentation model to enhance the global consistency of segmentation. In general, the optimization of the higher-order energy is extremely difficult. To make the problem tractable, we decompose the higher-order energy into two parts by utilizing auxiliary variables and then solve it by using an iterative process. We conduct quantitative and qualitative analyses on multiple datasets, and the proposed method achieves competitive results.

• Nuclear Engineering and Technology
• /
• 제53권12호
• /
• pp.4072-4079
• /
• 2021

This article presents a study on the state-of-the-art methods for automated radioactive material detection and identification, using gamma-ray spectra and modern machine learning methods. The recent developments inspired this in deep learning algorithms, and the proposed method provided better performance than the current state-of-the-art models. Machine learning models such as: fully connected, recurrent, convolutional, and gradient boosted decision trees, are applied under a wide variety of testing conditions, and their advantage and disadvantage are discussed. Furthermore, a hybrid model is developed by combining the fully-connected and convolutional neural network, which shows the best performance among the different machine learning models. These improvements are represented by the model's test performance metric (i.e., F1 score) of 93.33% with an improvement of 2%-12% than the state-of-the-art model at various conditions. The experimental results show that fusion of classical neural networks and modern deep learning architecture is a suitable choice for interpreting gamma spectra data where real-time and remote detection is necessary.