• Annual Conference of KIPS
• /
• 2017.11a
• /
• pp.835-838
• /
• 2017

최근 다양한 네트워크 서비스와 응용들이 생겨나면서, 네트워크상에 다양한 네트워크 트래픽이 발생하고 있다. 이로 인하여, 네트워크에 불필요한 네트워크 트래픽도 많이 발생하면서 네트워크 성능에 저하를 발생 시키고 있다. 따라서, 네트워크 트래픽 분류를 통하여 빠르게 제공되어야 하는 네트워크 서비스를 빠르게 전송 할 수 있도록 각 네트워크 트래픽마다의 분류가 필요하다. 본 논문에서는 Deep Learning 기법 중 Recurrent Neural Network를 이용한 플로우 기반의 네트워크 트래픽 분류를 제안한다. Deep Learning은 네트워크 관리자의 개입 없이 네트워크 트래픽 분류를 할 수 있으며, 이를 위하여 네트워크 트래픽을 Recurrent Neural Network에 적합한 데이터 형태로 변환한다. 변환된 데이터 세트를 이용하여 훈련시킴으로써 네트워크 트래픽을 분류한다. 본 논문에서는 훈련시킨 결과를 토대로 비교 분석 및 평가를 진행한다.

• ETRI Journal
• /
• v.35 no.2
• /
• pp.207-217
• /
• 2013

This paper introduces the modular cellular neural network (CNN), which is a new CNN structure constructed from nine one-layer modules with intercellular interactions between different modules. The new network is suitable for implementing many image processing operations. Inputting an image into the modules results in nine outputs. The topographic characteristic of the cell interactions allows the outputs to introduce new properties for image processing tasks. The stability of the system is proven and the performance is evaluated in several image processing applications. Experiment results on texture segmentation show the power of the proposed structure. The performance of the structure in a real edge detection application using the Berkeley dataset BSDS300 is also evaluated.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.10
• /
• pp.90-97
• /
• 1996

This paper presents an efficient parallel implementation and its performance evaluations of a SOFM neural netowrk on the multicomputer. We investigate the parallel performance as the size of a neural network N, the number of the patterns L, and the number of the processors p increase. We propose an analytica performance evaluation model for eac of the parallel implementations and verified the validity of the model through experiments. Analytical result show that the number of processors for a maximum speedup of the network decomposition nd the training-set decomposition increases in proportion to .root.N and .root.L, respectively. The performances of the both decompositions depend on the number of training patterns L and the size of the neural network N and, if L.geq.0.423N, the performance of trhe training-set decomposition is proved to be better than that of the network decomposition.

• Proceedings of the KIEE Conference
• /
• 1993.07a
• /
• pp.231-234
• /
• 1993

This paper deals with the design of an adaptive controller using neural network. We present RBFMLP Neural Network which consists of serial-connected two networks - Radial Basis Function Network and Multi Layer Perceptron, and then design a controller based on proposed networks with the adaptive control system structure, The plant and parameters of the controller are identified by the neural networks. We use the dynamic backpropagation algorithm for the learning of networks. Simulations represent the superiorities of the proposed network and the controller.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.2 no.1
• /
• pp.39-45
• /
• 1997

In this paper, a high speed motor drive system using an indirect adaptive neural network controller is proposed. In the variable high speed motor drives, the speed response can be deteriorated by long settling time and high overshoot. To obtain a good dynamical performance, an adaptive feedforward controller consisted of Neural Network Controller(NNC) and Neural Network Emulator(NNE) is applied. The NNE is used to identify the parameters and characteristics of high speed motor. To train the controller, the weights are dynamically adjusted using the back propagation algorithm. Computer simulation and implementation of the proposed system is described.

• Proceedings of the KIEE Conference
• /
• 2003.11c
• /
• pp.421-424
• /
• 2003

This paper proposes the generalized predictive control(GPC) method of chaotic systems using a self-recurrent wavelet neural network(SRWNN). The reposed SRWNN, a modified model of a wavelet neural network(WNN), has the attractive ability such as dynamic attractor, information storage for later use. Unlike a WNN, since the SRWNN has the mother wavelet layer which is composed of self-feedback neurons, mother wavelet nodes of the SRWNN can store the past information of the network. Thus the SRWNN can be used as a good tool for predicting the dynamic property of nonlinear dynamic systems. In our method, the gradient-descent(GD) method is used to train the SRWNN structure. Finally, the effectiveness and feasibility of the SRWNN based GPC is demonstrated with applications to a chaotic system.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.08a
• /
• pp.217-220
• /
• 2000

This paper describes the application of a neural network based adaptive line enhancer (ALE) for enhancement of the weak QRS complex corrupted with background noise. Modified fully-connected recurrent neural network is used as a nonlinear adaptive filter in the ALE. The connecting weights between network nodes as well as the parameters of the node activation function are updated at each iteration using the gradient descent algorithm. The real ECG signal buried with moderate and severe background noise is applied to the ALE. Simulation results show that the neural network based ALE performs well the enhancement of the QRS complex from noisy ECG signals.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.16 no.7
• /
• pp.610-614
• /
• 1991

This paper proposes a trajectory constrol fo a robot manipulator by using neural network. The inverse dynamic model of manipuator is learned by neural network. The manipulator is controlled by weight values of the learned neural network. The weight valuese is change with a torque of liner vontroller and a acceleration error. Phsically, the totlal torque for a manipualator is a sum of the liner controller torque and the nerural network controller torque. The proposed control effect is estimated by computer simulation.

• Journal of Information Processing Systems
• /
• v.17 no.2
• /
• pp.385-398
• /
• 2021

The complexity of deep learning models affects the real-time performance of gesture recognition, thereby limiting the application of gesture recognition algorithms in actual scenarios. Hence, a residual learning neural network based on a deep convolutional neural network is proposed. First, small convolution kernels are used to extract the local details of gesture images. Subsequently, a shallow residual structure is built to share weights, thereby avoiding gradient disappearance or gradient explosion as the network layer deepens; consequently, the difficulty of model optimisation is simplified. Additional convolutional neural networks are used to accelerate the refinement of deep abstract features based on the spatial importance of the gesture feature distribution. Finally, a fully connected cascade softmax classifier is used to complete the gesture recognition. Compared with the dense connection multiplexing feature information network, the proposed algorithm is optimised in feature multiplexing to avoid performance fluctuations caused by feature redundancy. Experimental results from the ISOGD gesture dataset and Gesture dataset prove that the proposed algorithm affords a fast convergence speed and high accuracy.

• Journal of the Semiconductor & Display Technology
• /
• v.20 no.1
• /
• pp.64-69
• /
• 2021

In this paper, a method for improving the defect classification performance in low contrast, ununiformity and featureless steel plate surfaces has been studied based on deep convolution neural network and transfer-learning neural network. The steel plate surface images have low contrast, ununiformity, and featureless, so that the contrast between defect and defect-free regions are not discriminated. These characteristics make it difficult to extract the feature of the surface defect image. A classifier based on a deep convolution neural network is constructed to extract features automatically for effective classification of images with these characteristics. As results of the experiment, AlexNet-based transfer-learning classifier showed excellent classification performance of 99.43% with less than 160 seconds of training time. The proposed classification system showed excellent classification performance for low contrast, ununiformity, and featureless surface images.