• Journal of Korea Water Resources Association
• /
• v.36 no.2
• /
• pp.195-209
• /
• 2003

Spatial-Stochastic Neural Networks Model(SSNNM) is used to estimate long-term streamflow in the parallel reservoir groups. SSNNM employs two kinds of backpropagation algorithms, based on LMBP and BFGS-QNBP separately. SSNNM has three layers, input, hidden, and output layer, in the structure and network configuration consists of 8-8-2 nodes one by one. Nodes in input layer are composed of streamflow, precipitation, pan evaporation, and temperature with the monthly average values collected from Andong and Imha reservoir. But some temporal differences apparently exist in their time series. For the SSNNM training procedure, the training sets in input layer are generated by the PARMA(1,1) stochastic model and they covers insufficient time series. Generated data series are used to train SSNNM and the model parameters, optimal connection weights and biases, are estimated during training procedure. They are applied to evaluate model validation using observed data sets. In this study, the new approaches give outstanding results by the comparison of statistical analysis and hydrographs in the model validation. SSNNM will help to manage and control water distribution and give basic data to develop long-term coupled operation system in parallel reservoir groups of the Upper Nakdong River.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.10
• /
• pp.4929-4947
• /
• 2017

In visual object tracking, Correlation Filter-based Tracking (CFT) systems have arouse recently to be the most accurate and efficient methods. The CFT's circularly shifts the larger search window to find most likely position of the target. The need of larger search window to cover both background and object make an algorithm sensitive to the background and the target occlusions. Further, the use of fixed-sized windows for training makes them incapable to handle scale variations during tracking. To address these problems, we propose two layer target representation in which both global and local appearances of the target is considered. Multiple local patches in the local layer provide robustness to the background changes and the target occlusion. The target representation is enhanced by employing additional reversed RGB channels to prevent the loss of black objects in background during tracking. The final target position is obtained by the adaptive weighted average of confidence maps from global and local layers. Furthermore, the target scale variation in tracking is handled by the statistical model, which is governed by adaptive constraints to ensure reliability and accuracy in scale estimation. The proposed structural enhancement is tested on VTBv1.0 benchmark for its accuracy and robustness.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.7
• /
• pp.173-182
• /
• 1996

In this paper, we propose the parallel, self-organizing, hierarchical neural netowrks as a handwritten digit recognition system. This system can absorb the various shape variations of handwritten digits by using the different methods of extracting the features in each stage neural network (SNN) of the PSHNN, and can reduce training time by using the single layer neural network as the SNN, and can obtain high rate of correct recognition by using the certainty area in all the output nodes individually. experiments have been performed with NIST database. In which we use 21, 315 digits (10, 625 digits for training and 10,663 digits for testing). The results show that the correct rate is 97.48% the error rate is 1.72% and the reject rate is 0.78%.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.11
• /
• pp.5568-5587
• /
• 2018

This research uses artificial intelligence methods for computer network intrusion detection system modeling. Primary classification is done using self-organized maps (SOM) in two levels, while the secondary classification of ambiguous data is done using Sugeno type Fuzzy Inference System (FIS). FIS is created by using Adaptive Neuro-Fuzzy Inference System (ANFIS). The main challenge for this system was to successfully detect attacks that are either unknown or that are represented by very small percentage of samples in training dataset. Improved algorithm for SOMs in second layer and for the FIS creation is developed for this purpose. Number of clusters in the second SOM layer is optimized by using our improved algorithm to minimize amount of ambiguous data forwarded to FIS. FIS is created using ANFIS that was built on ambiguous training dataset clustered by another SOM (which size is determined dynamically). Proposed hybrid model is created and tested using NSL KDD dataset. For our research, NSL KDD is especially interesting in terms of class distribution (overlapping). Objectives of this research were: to successfully detect intrusions represented in data with small percentage of the total traffic during early detection stages, to successfully deal with overlapping data (separate ambiguous data), to maximize detection rate (DR) and minimize false alarm rate (FAR). Proposed hybrid model with test data achieved acceptable DR value 0.8883 and FAR value 0.2415. The objectives were successfully achieved as it is presented (compared with the similar researches on NSL KDD dataset). Proposed model can be used not only in further research related to this domain, but also in other research areas.

• Smart Structures and Systems
• /
• v.33 no.1
• /
• pp.65-91
• /
• 2024

The use of five optimization techniques for the prediction of a strength-based concrete mixture's best-fit model is examined in this work. Five optimization techniques are utilized for this purpose: Slime Mold Algorithm (SMA), Black Hole Algorithm (BHA), Multi-Verse Optimizer (MVO), Vortex Search (VS), and Whale Optimization Algorithm (WOA). MATLAB employs a hybrid learning strategy to train an artificial neural network that combines least square estimation with backpropagation. Thus, 72 samples are utilized as training datasets and 31 as testing datasets, totaling 103. The multi-layer perceptron (MLP) is used to analyze all data, and results are verified by comparison. For training datasets in the best-fit models of SMA-MLP, BHA-MLP, MVO-MLP, VS-MLP, and WOA-MLP, the statistical indices of coefficient of determination (R²) in training phase are 0.9603, 0.9679, 0.9827, 0.9841 and 0.9770, and in testing phase are 0.9567, 0.9552, 0.9594, 0.9888 and 0.9695 respectively. In addition, the best-fit structures for training for SMA, BHA, MVO, VS, and WOA (all combined with multilayer perceptron, MLP) are achieved when the term population size was modified to 450, 500, 250, 150, and 500, respectively. Among all the suggested options, VS could offer a stronger prediction network for training MLP.

• Journal of Electrical Engineering and Technology
• /
• v.12 no.3
• /
• pp.1292-1298
• /
• 2017

In this paper, we represent a nonlinear identification of electronic brake pedal behavior in the brake-by-wire (BBW) system based on hybrid group method of data handling (GMDH) and genetic algorithm (GA). A GMDH is a kind of multi-layer network with a structure that is determined through training and which can express nonlinear dynamics as a mathematical model. The GA is used in the GMDH, enabling each neuron to search for its optimal set of connections with the preceding layer. The results obtained with this hybrid approach were compared with different nonlinear system identification methods. The experimental results showed that the hybrid approach performs better than the other methods in terms of root mean square error (RMSE) and correlation coefficients. The hybrid GMDH/GA approach was effective for modeling and predicting the brake pedal system under random braking conditions.

• Proceedings of the KIEE Conference
• /
• 2001.07d
• /
• pp.2726-2728
• /
• 2001

This paper describes an optimal synthesis method of binary neural network(BNN) for an approximation problem of a circular region using a newly proposed learning algorithm[7] Our object is to minimize the number of connections and neurons in hidden layer by using a Newly Expanded and Truncated Learning Algorithm(NETLA) for the multilayer BNN. The synthesis method in the NETLA is based on the extension principle of Expanded and Truncated Learning(ETL) and is based on Expanded Sum of Product (ESP) as one of the boolean expression techniques. And it has an ability to optimize the given BNN in the binary space without any iterative training as the conventional Error Back Propagation(EBP) algorithm[6] If all the true and false patterns are only given, the connection weights and the threshold values can be immediately determined by an optimal synthesis method of the NETLA without any tedious learning. Futhermore, the number of the required neurons in hidden layer can be reduced and the fast learning of BNN can be realized. The superiority of this NETLA to other algorithms was proved by the approximation problem of one circular region.

• Healthcare Informatics Research
• /
• v.24 no.4
• /
• pp.335-345
• /
• 2018

Objectives: The retinal nerve fiber layer (RNFL) is a site of glaucomatous optic neuropathy whose early changes need to be detected because glaucoma is one of the most common causes of blindness. This paper proposes an automated RNFL detection method based on the texture feature by forming a co-occurrence matrix and a backpropagation neural network as the classifier. Methods: We propose two texture features, namely, correlation and autocorrelation based on a co-occurrence matrix. Those features are selected by using a correlation feature selection method. Then the backpropagation neural network is applied as the classifier to implement RNFL detection in a retinal fundus image. Results: We used 40 retinal fundus images as testing data and 160 sub-images (80 showing a normal RNFL and 80 showing RNFL loss) as training data to evaluate the performance of our proposed method. Overall, this work achieved an accuracy of 94.52%. Conclusions: Our results demonstrated that the proposed method achieved a high accuracy, which indicates good performance.

• Journal of Digital Contents Society
• /
• v.11 no.4
• /
• pp.417-424
• /
• 2010

In this paper the real-time face region was detected by suggesting the rectangular feature-based classifier and the robust detection algorithm that satisfied the efficiency of computation and detection performance was suggested. By using the detected face region as a recognition input image, in this paper the face recognition method combined with PCA and the multi-layer network which is one of the intelligent classification was suggested and its performance was evaluated. As a pre-processing algorithm of input face image, this method computes the eigenface through PCA and expresses the training images with it as a fundamental vector. Each image takes the set of weights for the fundamental vector as a feature vector and it reduces the dimension of image at the same time, and then the face recognition is performed by inputting the multi-layer neural network.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.8
• /
• pp.1743-1750
• /
• 2003

DC motor requires a rotor speed sensor for accurate speed control. The speed sensors such as resolvers and encoders are used as a speed detector, but they increase cost and size of the motor and restrict the industrial drive applications. So in these days, many papers have reported in the sensorless operation of DC motor〔3­5〕. This paper presents a new sensorless strategy using neural networks〔6­8〕. Neural network has three layers which are input layer, hidden layer and output layer. The optimal neural network structure was tracked down by trial and error, and it was found that 4­16­1 neural network structure has given suitable results for the instantaneous rotor speed. Also, learning method is very important in neural network. Supervised learning methods〔8〕 are typically used to train the neural network for learning the input/output pattern presented. The back­propagation technique adjusts the neural network weights during training. The rotor speed is gained by weights and four inputs to the neural network. The experimental results were found satisfactory in both the independency on machine parameters and the insensitivity to the load condition.