• 한국멀티미디어학회논문지
• /
• 제25권10호
• /
• pp.1375-1385
• /
• 2022

In this paper, we propose a semi-supervised domain adaptation solution to deal with practical face recognition (FR) scenarios where a single face image for each target identity (to be recognized) is only available in the training phase. Main goal of the proposed method is to reduce the discrepancy between the target and the source domain face images, which ultimately improves FR performances. The proposed method is based on the Domain Adatation network (DAN) using an MMD loss function to reduce the discrepancy between domains. In order to train more effectively, we develop a novel loss function learning strategy in which MMD loss and cross-entropy loss functions are adopted by using different weights according to the progress of each epoch during the learning. The proposed weight adoptation focuses on the training of the source domain in the initial learning phase to learn facial feature information such as eyes, nose, and mouth. After the initial learning is completed, the resulting feature information is used to training a deep network using the target domain images. To evaluate the effectiveness of the proposed method, FR performances were evaluated with pretrained model trained only with CASIA-webface (source images) and fine-tuned model trained only with FERET's gallery (target images) under the same FR scenarios. The experimental results showed that the proposed semi-supervised domain adaptation can be improved by 24.78% compared to the pre-trained model and 28.42% compared to the fine-tuned model. In addition, the proposed method outperformed other state-of-the-arts domain adaptation approaches by 9.41%.

• 지능정보연구
• /
• 제7권2호
• /
• pp.143-151
• /
• 2001

이 논문은 서명의 분할 영역에 대한 자기회귀적(autoregressive) 특징을 사용하는 신경망 기반의 서명검증 시스템에 대해 다루고 있다. 이 논문에서 특기할 사항으로, 첫 번째, 서명 검증을 위한 신경망 구조의 설계와 학습 방법을 제시하고 있으며, 두 번째는 동일 서명자로부터 얻은 여러개의 서명에 대한 분할의 일관성을 위해서 DTW알고리듬을 적용한 것이다. 서명 검증 시스템의 성능분석을 위하여 정교하게 만들어진 방대한 양의 모조 서명 데이터베이스를 사용하였다. 1920개의 정교한 모조 서명에 대해 모조서명 인정율 0.78%, 진서명 거부율 1.6%의 결과를 보였다.

• 한국강구조학회 논문집
• /
• 제15권4호통권65호
• /
• pp.413-422
• /
• 2003

구조 설계를 위해 초기 부재를 가정할 때나 건축 실무에서 개산 견적을 계산할 때 기술자의 직관이나 비슷한 조건의 기존 설계 평균값을 사용하고 있으나 설계 조건은 모두 다르기 때문에 큰 오차가 발생할 수밖에 없다. 이러한 문제점을 해결하기 위해서 확률론적인 절차가 내재되어 있어 불확실성을 다룰 수 있는 인공 신경 회로망의 이용하여 와렌 트러스를 설계하므로써 적용성을 평가하였다. 제안된 신경망 설계변수값와 구조설계 단계에 따라 다양한 와렌 트러스를 설계하여 MIDAS 프로그램 설계결과의 10% 오차 이내로 근사 설계를 하므로써 모델의 타당성을 검증하였다. 제안된 모델은 약간의 오차를 포함하지만 적은 시간과 노력으로 신뢰할 수 있는 설계 결과를 얻을 수 있으며, 부재 테이블을 사용하는 비선형 관계의 구조설계에도 적용 가능한 특성을 가지고 있다.

• 한국소음진동공학회논문집
• /
• 제25권11호
• /
• pp.753-763
• /
• 2015

This paper proposes a decentralized robust adaptive neural network control scheme using multiple radial basis function neural networks for electrically driven robot manipulators with bounded input voltages in the presence of uncertainties. The proposed controller considers both robot link dynamics and actuator dynamics. Practically, the controller gain coefficients applied at each joint may be nonlinear time-varying and the input voltage at each joint is saturated. The proposed robot controller overcomes the various uncertainties and the input voltage saturation problem. The proposed controller does not require any robot and actuator parameters. The adaptation laws of the proposed controller are derived by using the Lyapunov stability analysis and the stability of the closed-loop control system is guaranteed. The validity and robustness of the proposed control scheme are verified through simulation results.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 2005년도 학술대회 논문집
• /
• pp.315-320
• /
• 2005

This paper is proposed adaptive fuzzy-neuro controller for high performance of induction motor drive. The design of this algorithm based on fuzzy-neural network controller that is implemented using fuzzy control and neural network. This controller uses fuzzy rule as training patterns of a neural network. Also, this controller uses the back-propagation method to adjust the weights between the neurons of neural network in order to minimize the error between the command output and actual output. A model reference adaptive scheme is proposed in which the adaptation mechanism is executed by fuzzy logic based on the error and change of nor measured between the motor speed and output of a reference model. The control performance of the adaptive fuzy-neuro controller is evaluated by analysis for various operating conditions. The results of experiment prove that the proposed control system has strong high performance and robustness to parameter variation, and steady-state accuracy and transient response.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 하계학술대회 논문집 B
• /
• pp.1248-1250
• /
• 2004

This paper is proposed adaptive fuzzy-neural network controller(NFC) for speed control of surface permanent magnet synchronous motor(SPMSM) drive. The design of this algorithm based on NFC that is implemented using fuzzy control and neural network. This controller uses fuzzy rule as training patterns of a neural network. Also, this controller uses the back-propagation method to adjust the weights between the neurons of neural network in order to minimize the error between the command output and actual output. A model reference adaptive scheme is proposed in which the adaptation mechanism is executed by fuzzy logic based on the error and change of error measured between the motor speed and output of a reference model. The control performance of the adaptive NFC is evaluated by analysis for various operating conditions. The results of analysis prove that the proposed control system has strong high performance and robustness to parameter variation, and steady-state accuracy and transient response.

• 전자통신동향분석
• /
• 제34권4호
• /
• pp.55-64
• /
• 2019

In this study, we introduce trends in neural-network-based deep learning research applied to dialogue systems. Recently, end-to-end trainable goal-oriented dialogue systems using long short-term memory, sequence-to-sequence models, among others, have been studied to overcome the difficulties of domain adaptation and error recognition and recovery in traditional pipeline goal-oriented dialogue systems. In addition, some research has been conducted on applying reinforcement learning to end-to-end trainable goal-oriented dialogue systems to learn dialogue strategies that do not appear in training corpora. Recent neural network models for end-to-end trainable chit-chat systems have been improved using dialogue context as well as personal and topic information to produce a more natural human conversation. Unlike previous studies that have applied different approaches to goal-oriented dialogue systems and chit-chat systems respectively, recent studies have attempted to apply end-to-end trainable approaches based on deep neural networks in common to them. Acquiring dialogue corpora for training is now necessary. Therefore, future research will focus on easily and cheaply acquiring dialogue corpora and training with small annotated dialogue corpora and/or large raw dialogues.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권10호
• /
• pp.4930-4951
• /
• 2018

The deep learning based speech enhancement has shown considerable success. However, it still suffers performance degradation under mismatch conditions. In this paper, an adaptation method is proposed to improve the performance under noise mismatch conditions. Firstly, we advise a noise aware training by supplying identity vectors (i-vectors) as parallel input features to adapt deep neural network (DNN) acoustic models with the target noise. Secondly, given a small amount of adaptation data, the noise-dependent DNN is obtained by using $L_2$ regularization from a noise-independent DNN, and forcing the estimated masks to be close to the unadapted condition. Finally, experiments were carried out on different noise and SNR conditions, and the proposed method has achieved significantly 0.1%-9.6% benefits of STOI, and provided consistent improvement in PESQ and segSNR against the baseline systems.

• Mass Spectrometry Letters
• /
• 제3권4호
• /
• pp.93-100
• /
• 2012

Cellular processes such as proliferation, differentiation, and adaptation to environmental changes are regulated by protein phosphorylation. In order to enhance the understanding of molecular dynamics for biological process in detail, it is necessary to develop sensitive and comprehensive analytical methods for the determination of protein phosphorylation. Neural stem cells hold great promise for neural repair following an injury or disease. In this study, we made differentiated oligodendrocytes from human neural stem cells using over-expression of olig2 gene. We confirmed using quantitative phosphoproteome analysis approach that combines stable isotope labeling by amino acids in cell culture (SILAC) and $TiO_2$ micro-column for phosphopeptide enrichment with $MS^2$ and $MS^3$ mass spectrometry. We detected 275 phosphopeptides which were modulated at least 2-fold between human neural stem cells and oligodendrocytes. Among them, 23 phosphoproteins were up-regulated in oligodendrocytes and 79 phosphoproteins were up-regulated in F3 cells.

• 한국융합학회논문지
• /
• 제4권3호
• /
• pp.43-50
• /
• 2013

This paper presents a model predictive control (MPC) based on a neural network (NN) model for air/fuel ration (AFR) control of automotive engines. The novelty of the paper is that the severe nonlinearity of the engine dynamics are modelled by a NN to a high precision, and adaptation of the NN model can cope with system uncertainty and time varying effects. A single dimensional optimization algorithm is used in the paper to speed up the optimization so that it can be implemented to the engine fast dynamics. Simulations on a widely used mean value engine model (MVEM) demonstrate effectiveness of the developed method.