• 제어로봇시스템학회논문지
• /
• 제21권1호
• /
• pp.59-64
• /
• 2015

In this paper, we found the usefulness of the deep belief network (DBN) in the fields of brain-computer interface (BCI), especially in relation to imagined speech. In recent years, the growth of interest in the BCI field has led to the development of a number of useful applications, such as robot control, game interfaces, exoskeleton limbs, and so on. However, while imagined speech, which could be used for communication or military purpose devices, is one of the most exciting BCI applications, there are some problems in implementing the system. In the previous paper, we already handled some of the issues of imagined speech when using the International Phonetic Alphabet (IPA), although it required complementation for multi class classification problems. In view of this point, this paper could provide a suitable solution for vowel classification for imagined speech. We used the DBN algorithm, which is known as a deep learning algorithm for multi-class vowel classification, and selected four vowel pronunciations:, /a/, /i/, /o/, /u/ from IPA. For the experiment, we obtained the required 32 channel raw electroencephalogram (EEG) data from three male subjects, and electrodes were placed on the scalp of the frontal lobe and both temporal lobes which are related to thinking and verbal function. Eigenvalues of the covariance matrix of the EEG data were used as the feature vector of each vowel. In the analysis, we provided the classification results of the back propagation artificial neural network (BP-ANN) for making a comparison with DBN. As a result, the classification results from the BP-ANN were 52.04%, and the DBN was 87.96%. This means the DBN showed 35.92% better classification results in multi class imagined speech classification. In addition, the DBN spent much less time in whole computation time. In conclusion, the DBN algorithm is efficient in BCI system implementation.

• 한국음향학회지
• /
• 제37권6호
• /
• pp.506-511
• /
• 2018

본 논문에서는 잡음 환경에서 효과적인 음성인식을 위하여 GMM(Gaussian Mixture Model)-DNN(Deep Neural Network) 하이브리드 기반의 특징 보상 기법을 제안한다. 기존의 GMM 기반의 특징 보상에서 필요로 하는 사후 확률을 DNN을 통해 계산한다. Aurora 2.0 데이터를 이용한 음성 인식 성능 평가에서 본 논문에서 제안한 GMM-DNN 하이브리드 기법이 기존의 GMM 기반 기법에 비해 Known, Unknown 잡음 환경에서 모두 평균적으로 우수한 성능을 나타낸다. 특히 Unknown 잡음 환경에서 평균 오류율이 9.13 %의 상대 향상률을 나타내고, 낮은 SNR(Signal to Noise Ratio) 잡음 환경에서 상당히 우수한 성능을 보인다.

• 한국정보통신학회논문지
• /
• 제26권3호
• /
• pp.430-435
• /
• 2022

본 논문에서는 하나의 매크로 기지국과 다수의 소형 기지국들로 구성된 이종 네트워크 (Heterogeneous Network, HetNET) 시스템에서 비직교 다중 접속 (Non-Orthogonal Multiple Access, NOMA) 기술을 고려한다. 여기서, NOMA 신호에 대하여 완벽한 순차적 간접 제거를 가정한다. 본 논문에서는 이러한 NOMA 기반의 이종 네트워크에서 데이터 전송률을 최대화하기 위하여 딥러닝 기반의 사용자 및 전력 할당 기법을 제안한다. 특히, 제안하는 기법은 부하 분산을 위한 심층신경망(Deep Neural Network, DNN) 기반의 사용자 할당 과정과 할당된 사용자에 대한 데이터 전송률의 최대화를 위한 DNN 기반의 전력 할당 과정을 포함한다. 기지국과 사용자간 경로 손실과 레일레이 페이딩 채널을 가정한 시뮬레이션을 통해 제안하는 기법의 성능을 평가하고, 기존의 최대 신호 대 간섭 및 잡음비(Max-Signal-to-Interference-plus-Noise Ratio, Max-SINR) 기법의 성능과 비교한다. 성능 비교를 통해서 제안된 기법이 기존의 Max-SINR 기법보다 높은 데이터 전송률을 제공하는 것을 보여준다.

• 인터넷정보학회논문지
• /
• 제21권5호
• /
• pp.21-29
• /
• 2020

Accurate estimation of human pose relies on backbone method in which its role is to extract feature map. Up to dated, the method of backbone feature extraction is conducted by the plain convolutional neural networks named by CNN and the residual neural networks named by Resnet, both of which have various architectures and performances. The CNN family network such as VGG which is well-known as a multiple stacked hidden layers architecture of deep learning methods, is base and simple while Resnet which is a bottleneck layers architecture yields fewer parameters and outperform. They have achieved inspired results as a backbone network in human pose estimation. However, they were used then followed by different pose estimation networks named by pose parsing module. Therefore, in this paper, we present a comparison between the plain CNN family network (VGG) and bottleneck network (Resnet) as a backbone method in the same pose parsing module. We investigate their performances such as number of parameters, loss score, precision and recall. We experiment them in the bottom-up method of human pose estimation system by adapted the pose parsing module of openpose. Our experimental results show that the backbone method using VGG network outperforms the Resent network with fewer parameter, lower loss score and higher accuracy of precision and recall.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2019년도 학술발표회
• /
• pp.156-156
• /
• 2019

In this study, a gated recurrent unit (GRU) network is constructed based on a deep neural network (DNN) with the aim of restoring the missing daily flow data in river basins. Lai Chau hydrological station is located upstream of the Da river basin (Vietnam) is selected as the target station for this study. Input data of the model are data on observed daily flow for 24 years from 1961 to 1984 (before Hoa Binh dam was built) at 5 hydrological stations, in which 4 gauge stations in the basin downstream and restoring - target station (Lai Chau). The total available data is divided into sections for different purposes. The data set of 23 years (1961-1983) was employed for training and validation purposes, with corresponding rates of 80% for training and 20% for validation respectively. Another data set of one year (1984) was used for the testing purpose to objectively verify the performance and accuracy of the model. Though only a modest amount of input data is required and furthermore the Lai Chau hydrological station is located upstream of the Da River, the calculated results based on the suggested model are in satisfactory agreement with observed data, the Nash - Sutcliffe efficiency (NSE) is higher than 95%. The finding of this study illustrated the outstanding performance of the GRU network model in recovering the missing flow data at Lai Chau station. As a result, DNN models, as well as GRU network models, have great potential for application within the field of hydrology and hydraulics.

• Journal of Information Processing Systems
• /
• 제17권3호
• /
• pp.556-570
• /
• 2021

The existing video expression recognition methods mainly focus on the spatial feature extraction of video expression images, but tend to ignore the dynamic features of video sequences. To solve this problem, a multi-mode convolution neural network method is proposed to effectively improve the performance of facial expression recognition in video. Firstly, OpenFace 2.0 is used to detect face images in video, and two deep convolution neural networks are used to extract spatiotemporal expression features. Furthermore, spatial convolution neural network is used to extract the spatial information features of each static expression image, and the dynamic information feature is extracted from the optical flow information of multiple expression images based on temporal convolution neural network. Then, the spatiotemporal features learned by the two deep convolution neural networks are fused by multiplication. Finally, the fused features are input into support vector machine to realize the facial expression classification. Experimental results show that the recognition accuracy of the proposed method can reach 64.57% and 60.89%, respectively on RML and Baum-ls datasets. It is better than that of other contrast methods.

• International Journal of Internet, Broadcasting and Communication
• /
• 제14권1호
• /
• pp.53-63
• /
• 2022

In the paper, we propose a TSCH-based scheduling scheme for IEEE 802.15.4e, which is able to perform the scheduling of its own network by avoiding collision from interference network cluster (INC). Firstly, we model a bipartite graph structure for presenting the slot-frame (channel-slot assignment) of TSCH. Then, based on the bipartite graph edge weight, we utilize the Hungarian assignment algorithm to implement a scheduling scheme. We have employed two features (maximization and minimization) of the Hungarian-based assignment algorithm, which can perform the assignment in terms of minimizing the throughput of INC and maximizing the throughput of own network. Further, in this work, we called the scheme "dual-stage Hungarian-based assignment algorithm". Furthermore, we also propose deep learning (DL) based deep neural network (DNN)scheme, where the data were generated by the dual-stage Hungarian-based assignment algorithm. The performance of the DNN scheme is evaluated by simulations. The simulation results prove that the proposed DNN scheme providessimilar performance to the dual-stage Hungarian-based assignment algorithm while providing a low execution time.

• Smart Structures and Systems
• /
• 제32권3호
• /
• pp.179-193
• /
• 2023

This study explores an alternative to the existing centralized process for data anomaly detection in modern Internet of Things (IoT)-based structural health monitoring (SHM) systems. An edge intelligence framework is proposed for the early detection and classification of various data anomalies facilitating quality enhancement of acquired data before transmitting to a central system. State-of-the-art deep neural network pruning techniques are investigated and compared aiming to significantly reduce the network size so that it can run efficiently on resource-constrained edge devices such as wireless smart sensors. Further, depthwise separable convolution (DSC) is invoked, the integration of which with advanced structural pruning methods exhibited superior compression capability. Last but not least, quantization-aware training (QAT) is adopted for faster processing and lower memory and power consumption. The proposed edge intelligence framework will eventually lead to reduced network overload and latency. This will enable intelligent self-adaptation strategies to be employed to timely deal with a faulty sensor, minimizing the wasteful use of power, memory, and other resources in wireless smart sensors, increasing efficiency, and reducing maintenance costs for modern smart SHM systems. This study presents a theoretical foundation for the proposed framework, the validation of which through actual field trials is a scope for future work.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권8호
• /
• pp.2053-2067
• /
• 2023

This paper proposes two video quality assessment methods based on deep neural network. (i)The first method uses the IQF-CNN (convolution neural network based on image quality features) to build image quality assessment method. The LIVE image database is used to test this method, the experiment show that it is effective. Therefore, this method is extended to the video quality assessment. At first every image frame of video is predicted, next the relationship between different image frames are analyzed by the hysteresis function and different window function to improve the accuracy of video quality assessment. (ii)The second method proposes a video quality assessment method based on convolution neural network (CNN) and gated circular unit network (GRU). First, the spatial features of video frames are extracted using CNN network, next the temporal features of the video frame using GRU network. Finally the extracted temporal and spatial features are analyzed by full connection layer of CNN network to obtain the video quality assessment score. All the above proposed methods are verified on the video databases, and compared with other methods.

• 한국정보통신학회논문지
• /
• 제21권1호
• /
• pp.144-150
• /
• 2017

최근 데이터의 지능적 처리 및 정확도 향상을 위해 딥러닝 기술이 응용되고 있다. 이 기술은 다층의 데이터 처리 레이어들로 구성된 계산 모델을 통해 이루어지는데, 이 모델은 여러 수준의 추상화를 거쳐 데이터의 표현을 학습한다. 딥러닝의 한 부류인 컨볼루션 신경망은 인간 행동 추정, 얼굴 인식, 이미지 분류, 음성 인식 같은 연구 분야에서 많이 활용되고 있다. 이미지 분류에 좋은 성능을 보여주는 컨볼루션 신경망은 깊은 학습망과 많은 부류를 이용하면 효과적으로 분류율을 높일수 있지만, 적은 부류의 데이터를 사용할 경우, 과적합 문제가 발생할 확률이 높아진다. 따라서 본 논문에서는 컨볼루션 신경망기반의 소부류의 분류을 위한 학습망을 제작하여 자체적으로 구축한 이미지 DB를 학습시키고, 객체를 분류하는 연구를 실험 하였으며, 1000개의 부류를 분류하기 위해 제작된 기존 공개된 망들과 비교 실험을 통해 기존 망보다 평균 7.06%이상의 상승된 분류율을 보여주었다.