• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.527-529
• /
• 2023

Generative Adversarial Networks (GANs) models have developed rapidly due to the emergence of various variation models and their wide applications. Despite many recent developments in GANs, mode collapse, and instability are still unresolved issues. To address these problems, we focused on the fact that a single GANs model itself cannot realize local failure during the training phase without external standards. This paper introduces a novel training process involving multiple GANs, inspired by auction mechanisms. During the training, auxiliary performance metrics for each GANs are determined by the others through the process of various auction methods.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.12C
• /
• pp.1177-1183
• /
• 2009

In this paper, support vector machines (SVM) are applied to multi-user detector (MUD) for direct sequence (DS)-CDMA system. This work shows an analytical performance of SVM based multi-user detector with some of kernel functions, such as linear, sigmoid, and Gaussian. The basic idea in SVM based training is to select the proper number of support vectors by maximizing the margin between two different classes. In simulation studies, the performance of SVM based MUD with different kernel functions is compared in terms of the number of selected support vectors, their corresponding decision boundary, and finally the bit error rate. It was found that controlling parameter, in SVM training have an effect, in some degree, to SVM based MUD with both sigmoid and Gaussian kernel. It is shown that SVM based MUD with Gaussian kernels outperforms those with other kernels.

• IE interfaces
• /
• v.24 no.2
• /
• pp.97-104
• /
• 2011

Research for multi-variate classification has been studied through two kinds of procedures which are feature selection and classification. Feature Selection techniques have been applied to select important features and the other one has improved classification performances through classifier applications. In general, each technique has been independently studied, however consideration of the interaction between both procedures has not been widely explored which leads to a degraded performance. In this paper, through integrating these two procedures, classification performance can be improved. The proposed model takes advantage of KBANN (Knowledge-Based Artificial Neural Network) which uses prior knowledge to learn NN (Neural Network) as training information. Each NN learns characteristics of the Feature Selection and Classification techniques as training sets. The integrated NN can be learned again to modify features appropriately and enhance classification performance. This innovative technique is called ALBNN (Algorithm Learning-Based Neural Network). The experiments' results show improved performance in various classification problems.

• Proceedings of the IEEK Conference
• /
• 2003.07e
• /
• pp.1968-1971
• /
• 2003

We propose a knowledge-based algorithm for extracting an object boundary from low-quality image like the forward looking infrared image. With the multi-classes training data set, the global shape is modeled by multispace KL(MKL)[1] and curvature model. And the objective function for fitting the deformable boundary template represented by the shape model to true boundary in an input image is formulated by Bales rule. Simulation results show that our method has more accurateness in case of multi-classes training set and performs better in the sense of computation cost than point distribution model(PDM)[2]. It works well in distortion under the noise, pose variation and some kinds of occlusions.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.5
• /
• pp.515-522
• /
• 2021

It is hard to prepare sufficient training data for speech emotion recognition due to the difficulty of emotion labeling. In this paper, we apply transfer learning with large-scale training data for speech recognition on a transformer-based model to improve the performance of speech emotion recognition. In addition, we propose a method to utilize context information without decoding by multi-task learning with speech recognition. According to the speech emotion recognition experiments using the IEMOCAP dataset, our model achieves a weighted accuracy of 70.6 % and an unweighted accuracy of 71.6 %, which shows that the proposed method is effective in improving the performance of speech emotion recognition.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.1
• /
• pp.88-98
• /
• 1994

A new neural network architecture called the Parallel. Self-Organizing Hierarchical Neural Network (PSHNN) is presented. The new architecture involves a number of stages in which each stage can be a particular neural network (SNN). The experiments performed in comparison to multi-layered network with backpropagation training and indicated the superiority of the new architecture in the sense of classification accuracy, training time,parallelism.

• IE interfaces
• /
• v.10 no.1
• /
• pp.135-143
• /
• 1997

The importance of training for NC, CNC and Machining Center has been greatly increased. This paper presents implementation of a DNC(Direct Numerical Control) operating software for educational system. This system is able to connect 8-32 CNCs to Control PC with RS232 multi-port serial card. Therefore, it allows much efficiency in training even after costs are considered. The KISCO DNC S/W for above system includes various communication functions, communication parameters setting, program editor and user-friendly environment. This software was developed with C and Windows programming. It was proved in function and stability by iterative field tests.

• 제어로봇시스템학회:학술대회논문집
• /
• 1996.10b
• /
• pp.1328-1331
• /
• 1996

The importance of training for NC, CNC and Machining Center has been greatly increased. This paper presents implementation of a DNC system and operating software for educational purpose. This system is able to connect 8-32 CNCs to Control PC with RS232 multi-port serial card. Therefore, it allows much efficiency in training even after costs are considered. The KISCO DNC S/W for above system includes various communication function, communication parameter setting, program editor, tool management and user-friendly environment.

• The Journal of Korean Physical Therapy
• /
• v.32 no.3
• /
• pp.185-190
• /
• 2020

Purpose: Gait and cognitive impairment in stroke patients exacerbate fall risk and mobility difficulties during multi-task walking. Virtual reality can provide interesting and challenging training in a community setting. This study evaluated the effect of community-based virtual reality gait training (VRGT) using a 360-degree image on the gait ability of chronic stroke patients. Methods: Forty-five chronic stroke patients who were admitted to a rehabilitation hospital participated in this study. Patients meeting the selection criteria were randomly divided into a VRGT group (n=23) and a control group (n=22). Both these groups received general rehabilitation. The VRGT group was evaluated using a 360-degree image that was recorded for 50 minutes a day, 5 days per week for a total of 6 weeks after their training. The control group received general treadmill training for the same amount of time as that of the VRGT group. The improvement in the spatiotemporal parameters of gait was evaluated using a gait analyzer system before and after training. Results: The spatiotemporal gait parameters showed significant improvements in both groups compare with the baseline measurements (p<0.05), and the VRGT group showed more improvement than the control group (p<0.05). Conclusion: Community-based VRGT has been shown to improve the walking ability of chronic stroke patients and is expected to be used in rehabilitation of stroke patients in the future.

• Journal of Broadcast Engineering
• /
• v.26 no.7
• /
• pp.855-861
• /
• 2021

Neural Architecture Search (NAS) is cutting-edge technology in the machine learning community. NAS Without Training (NASWOT) recently has been proposed to tackle the high demand of computational resources in NAS by leveraging some indicators to predict the performance of architectures before training. The advantage of these indicators is that they do not require any training. Thus, NASWOT reduces the searching time and computational cost significantly. However, NASWOT only considers high-performing networks which does not guarantee a fast inference speed on hardware devices. In this paper, we propose a multi objectives reward function, which considers the network's latency and the predicted performance, and incorporate it into the Reinforcement Learning approach to search for the best networks with low latency. Unlike other methods, which use FLOPs to measure the latency that does not reflect the actual latency, we obtain the network's latency from the hardware NAS bench. We conduct extensive experiments on NAS-Bench-201 using CIFAR-10, CIFAR-100, and ImageNet-16-120 datasets, and show that the proposed method is capable of generating the best network under latency constrained without training subnetworks.