• Journal of the Korean Operations Research and Management Science Society
• /
• v.27 no.1
• /
• pp.33-51
• /
• 2002

A fully connected Artificial Neural Network (ANN) contains many connections. Compared to the pruned ANN with fewer connections, the fully connected ANN takes longer time to produce solutions end may not provide appropriate solutions to new unseen date. Therefore, by reducing the sloe of ANN, we can overcome the overfitting problem and increase the computing speed. In this research, we reduced the size of ANN by destroying the connections. In other words, we investigated the performance change of the reduced ANN by systematically destroying the connections. Then we found the acceptable level of connection-destruction on which the resulting ANN Performs as well as the original fully connected ANN. In the previous researches on the sloe reduction of ANN, the reduced ANN had to be retrained every time some connections were eliminated. Therefore, It tool lolly time to obtain the reduced ANN. In this research, however, we provide the acceptable level of connection-destruction according to the size of the fully connected ANN. Therefore, by applying the acceptable level of connection-destruction to the fully connected ANN without any retraining, the reduced ANN can be obtained efficiently.

• Steel and Composite Structures
• /
• v.36 no.5
• /
• pp.493-506
• /
• 2020

The load-slip relationship of the shear connection is an important parameter in design and analysis of composite structures. In this paper, a load-slip curve prediction method of the shear connection based on the artificial neural networks (ANNs) is proposed. The factors which are significantly related to the structural and deformation performance of the connection are selected, and the shear stiffness of shear connections and the transverse coordinate slip value of the load-slip curve are taken as the input parameters of the network. Load values corresponding to the slip values are used as the output parameter. A twolayer hidden layer network with 15 nodes and 10 nodes is designed. The test data of two different forms of shear connections, the stud shear connection and the perforated shear connection with flange heads, are collected from the previous literatures, and the data of six specimens are selected as the two prediction data sets, while the data of other specimens are used to train the neural networks. Two trained networks are used to predict the load-slip curves of their corresponding prediction data sets, and the ratio method is used to study the proximity between the prediction loads and the test loads. Results show that the load-slip curves predicted by the networks agree well with the test curves.

• Journal of Korea Society of Industrial Information Systems
• /
• v.22 no.5
• /
• pp.17-24
• /
• 2017

Recently Deep-learning, Technologies using Large or Deep Artificial Neural Networks, have Shown Remarkable Performance, and the Increasing Size of the Network Contributes to its Performance Improvement. However, the Increase in the Size of the Neural Network Leads to an Increase in the Calculation Amount, which Causes Problems Such as Circuit Complexity, Price, Heat Generation, and Real-time Restriction. In This Paper, We Propose and Test a Method to Reduce the Number of Network Connections by Effectively Pruning the Redundancy in the Connection and Showing the Difference between the Performance and the Desired Range of the Original Neural Network. In Particular, we Proposed a Simple Method to Improve the Performance by Re-learning and to Guarantee the Desired Performance by Allocating the Error Rate per Layer in Order to Consider the Difference of each Layer. Experiments have been Performed on a Typical Neural Network Structure such as FCN (full connection network) and CNN (convolution neural network) Structure and Confirmed that the Performance Similar to that of the Original Neural Network can be Obtained by Only about 1/10 Connection.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.635-639
• /
• 2004

In this paper, we present a novel approach for the structure of Fuzzy Neural Network(FNN) based on wavelet function and apply this network structure to the modeling of chaotic nonlinear systems. Generally, the wavelet fuzzy model(WFM) has the advantage of the wavelet transform by constituting the fuzzy basis function(FBF) and the conclusion part to equalize the linear combination of FBF with the linear combination of wavelet functions. However, it is very difficult to identify the fuzzy rules and to tune the membership functions of the fuzzy reasoning mechanism. Neural networks, on the other hand, utilize their learning capability for automatic identification and tuning. Therefore, we design a wavelet based FNN structure(WFNN) that merges these advantages of neural network, fuzzy model and wavelet transform. The basic idea of our wavelet based FNN is to realize the process of fuzzy reasoning of wavelet fuzzy system by the structure of a neural network and to make the parameters of fuzzy reasoning be expressed by the connection weights of a neural network. And our network can automatically identify the fuzzy rules by modifying the connection weights of the networks via the gradient descent scheme. To verify the efficiency of our network structure, we evaluate the modeling performance for chaotic nonlinear systems and compare it with those of the FNN and the WFM.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.4 no.1
• /
• pp.111-118
• /
• 2004

In this paper, we present a novel approach for the structure of Fuzzy Neural Network(FNN) based on wavelet function and apply this network structure to the solution of the tracking problem for mobile robots. Generally, the wavelet fuzzy model(WFM) has the advantage of the wavelet transform by constituting the fuzzy basis function(FBF) and the conclusion part to equalize the linear combination of FBF with the linear combination of wavelet functions. However, it is very difficult to identify the fuzzy rules and to tune the membership functions of the fuzzy reasoning mechanism. Neural networks, on the other hand, utilize their learning capability for automatic identification and tuning. Therefore, we design a wavelet based FNN structure(WFNN) that merges these advantages of neural network, fuzzy model and wavelet transform. The basic idea of our wavelet based FNN is to realize the process of fuzzy reasoning of wavelet fuzzy system by the structure of a neural network and to make the parameters of fuzzy reasoning be expressed by the connection weights of a neural network. And our network can automatically identify the fuzzy rules by modifying the connection weights of the networks via the gradient descent scheme. To verify the efficiency of our network structure, we evaluate the tracking performance for mobile robot and compare it with those of the FNN and the WFM.

• Proceedings of the KIEE Conference
• /
• 1995.07b
• /
• pp.855-858
• /
• 1995

A neural network has very simple construction (input, output and connection weight) and then it can be robusted against some disturbance. In this paper, we proposed a neuro-controller using a Multi-Layered neural network which is combined with PD controller. The proposed neuro-controller is learned by backpropagation learning rule with momentum and neuro-controller adjusts connection weight in neural network to make approximate dynamic model of DC Servo motor. Computer Simulation results show that the proposed neuro-controller's performance is better than that of origianl PD controller.

• 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.

• Proceedings of the IEEK Conference
• /
• 1998.06a
• /
• pp.909-912
• /
• 1998

In this paper, a new connection admission controller using neural network is presented. The controller measures traffic flow, cell loss rate, and cell delay periodically. Using those measured information, it learns the distributions of traffics of each traffic. Also the proposed controller is able to measure and manage the delays that source traffics experience through the network by using DWRR multiplexer with buffers dedicated to each traffic source. Experimental result show that the heterogeneous traffic sources with various QoS requirement.

• 제어로봇시스템학회:학술대회논문집
• /
• 1989.10a
• /
• pp.124-126
• /
• 1989

Inverse kinematic problem is a crucial point for robot manipulator control. In this paper, to implement the Jacobian control technique we used the Hopfield(Tank)'s neural network. The states of neurons represent joint veocities, and the connection weights are determined from the current value of the Jacobian matrix. The network energy function is constructed so that its minimum corresponds to the minimum least square error. At each sampling time, connection weights and neuron states are updated according to current joint position. Inverse kinematic solution to the planar redundant manipulator is solved by computer simulation.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.65 no.6
• /
• pp.1070-1079
• /
• 2016

In this paper, the design of recursive radial basis function neural networks based on incremental fuzzy c-means is introduced for processing the big data. Radial basis function neural networks consist of condition, conclusion and inference phase. Gaussian function is generally used as the activation function of the condition phase, but in this study, incremental fuzzy clustering is considered for the activation function of radial basis function neural networks, which could effectively do big data processing. In the conclusion phase, the connection weights of networks are given as the linear function. And then the connection weights are calculated by recursive least square estimation. In the inference phase, a final output is obtained by fuzzy inference method. Machine Learning datasets are employed to demonstrate the superiority of the proposed classifier, and their results are described from the viewpoint of the algorithm complexity and performance index.