• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.21 no.10
• /
• pp.1561-1570
• /
• 1997

This study explores the treatment of the max-value cost function over a parameter interval in parametric optimization. To avoid the computational burden of the transformation treatment using an artificial variable, a direct treatment of the original max-value cost function is proposed. It is theoretically shown that the transformation treatment results in demanding an additional equality constraint of dual variables as a part of the Kuhn-Tucker necessary conditions. Also, it is demonstrated that the usability and feasibility conditions on the search direction of the transformation treatment retard convergence rate. To investigate numerical performances of both treatments, typical optimization algorithms in ADS are employed to solve a min-max steady-state response optimization. All the algorithm tested reveal that the suggested direct treatment is more efficient and stable than the transformation treatment. Also, the better performing of the direct treatment over the transformation treatment is clearly shown by constrasting the convergence paths in the design space of the sample problem. Six min-max transient response optimization problems are also solved by using both treatments, and the comparisons of the results confirm that the performances of the direct treatment is better than those of the tranformation treatment.

• Journal of Electrical Engineering and Technology
• /
• v.13 no.6
• /
• pp.2468-2478
• /
• 2018

We propose the Sparse Feature Convolutional Neural Network (SFCNN) to reduce the volume of convolutional neural networks (CNNs). Despite the superior classification performance of CNNs, their enormous network volume requires high computational cost and long processing time, making real-time applications such as online-training difficult. We propose an advanced network that reduces the volume of conventional CNNs by producing a region-based sparse feature map. To produce the sparse feature map, two complementary region-based value extraction methods, cluster max extraction and local value extraction, are proposed. Cluster max is selected as the main function based on experimental results. To evaluate SFCNN, we conduct an experiment with two conventional CNNs. The network trains 59 times faster and tests 81 times faster than the VGG network, with a 1.2% loss of accuracy in multi-class classification using the Caltech101 dataset. In vehicle classification using the GTI Vehicle Image Database, the network trains 88 times faster and tests 94 times faster than the conventional CNNs, with a 0.1% loss of accuracy.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.12 no.10
• /
• pp.1875-1882
• /
• 2008

Nowadays, usage of mobile unit which has a characteristic of low cost and high efficiency is being generalized because of frequent use of internet-based variable service and application in IEEE 802.16 WiMAX. A study for handling a security problem of high speed internet service is rising while the use of a mobile is being generalized. This paper suggests a security mechanism which provides safety from certification load of SS and a security attack as well as a basic function which is provided from IEEE 802.16e standard to satisfy security demand of IEEE802.16 WiMAX. The proposed mechanism exchangeskey material information for TEK and data code by using 난수(?) and secret value created by SS and BS, also reduces capacity load of BS not to perform an additional certificate procedure of BS by using the early certification information and certificate of SS.