• 대한임베디드공학회논문지
• /
• 제16권1호
• /
• pp.1-8
• /
• 2021

Detecting Out-of-Distribution (OOD) data is fundamentally required when Deep Neural Network (DNN) is applied to real-world AI such as autonomous driving. However, modern DNNs are quite vulnerable to the over-confidence problem even if the test data are far away from the trained data distribution. To solve the problem, this paper proposes a novel Adversarial-Mixup training method to let the DNN model be more robust by detecting OOD data effectively. Experimental results show that the proposed Adversarial-Mixup method improves the overall performance of OOD detection by 78% comparing with the State-of-the-Art methods. Furthermore, we show that the proposed method can alleviate the over-confidence problem by reducing the confidence score of OOD data than the previous methods, resulting in more reliable and robust DNNs.

• 유통과학연구
• /
• 제16권2호
• /
• pp.19-24
• /
• 2018

Purpose - Drone delivery is expected to revolutionize the supply chain industry. This paper aims to introduce a collaborative parcel delivery problem by truck and drone (hereinafter called "TDRP") and propose a novel heuristic method to solve the problem. Research design, data, and methodology - To show the effectiveness of collaborative delivery by truck and drone, we generate a toy problem composed of 9 customers and the speed of drone is assumed to be two times faster than truck. We compared the delivery completion times by 'truck only' case and 'truck and drone' case by solving the optimization problem respectively. Results - We provide literature reviews for truck and drone routing problem for collaborative delivery and propose a novel and original heuristic method to solve the problem with numerical example. By numerical example, collaborative delivery is expected to reduce delivery completion time by 12~33% than 'truck only' case. Conclusions - In this paper, we introduce the TDRP in order for collaborative delivery to be effective and propose a novel and original heuristic method to solve the problem. The results of research will be help to develop effective heuristic solution and optimize the parcel delivery by using drone.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1994년도 춘계공동학술대회논문집; 창원대학교; 08월 09일 Apr. 1994
• /
• pp.303-312
• /
• 1994

The problem of optimally allocating available communication channels in a cellular mobile system with nonuniform traffic distribution is considered. This problem is to minimize the weighted average blocking probability subject to cochannel interference constraints. We use the concept of pattern to deal with the problem more conveniently. Using Lagrangian relaxation and subgradient optimization techniques, we obtain high-quality solutions with information about their deviations from true optimal solutions. Computational experiments show that our method works very well.

• 대한전기학회논문지
• /
• 제40권2호
• /
• pp.135-143
• /
• 1991

The setting problem of protective devices in power distribution systems is a combinatorial search problem involving a lot of trial & errors to search for a proper solution which has a great impact for the system reliability. This paper discusses the nature of the problem and proposes a scheme to prune the decision tree achieving the efficient search for a setting solution by applying heuristic rules which utilize pattern informations of backup and primary devices. Its effectiveness is demonstrated through many tests on real systems.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2005년도 춘계학술대회 논문집 전기설비전문위원
• /
• pp.62-63
• /
• 2005

Recently, sectionalizing switches have been coming to be operated by remote control through the distribution SCADA system. However, the problem of determining the optimal switching sequence is a combinatorial optimization problem, and is quite difficult to solve, Hence, it is imperative to develop practically applicable solution algorithms create a new arbitral distribution system configuration from an initial configuration, and some of these algorithms do not show a load transfer sequence to reach the objective system.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 2004년도 학술대회 논문집
• /
• pp.381-383
• /
• 2004

Recently, sectionalizing switches have been coming to be operated by remote control through the distribution SCADA system. However, the problem of determining the optimal switching sequence is a combinatorial optimization problem, and is quite difficult to solve. Hence, it is imperative to develop practically applicable solution algorithms create a new arbitral distribution system configuration from an initial configuration, and some of these algorithms do not show a load transfer sequence to reach the objective system.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 2005년도 춘계학술대회논문집
• /
• pp.418-420
• /
• 2005

Recently, sectionalizing switches have been coming to be operated by remote control through the distribution SCADA system. However, the problem of determining the optimal switching sequence is a combinatorial optimization problem, and is quite difficult In solve. Hence, it is imperative to develop practically applicable solution algorithms create a new arbitral distribution system configuration from an initial configuration and some of these algorithms do not show a load transfer sequence to reach the objective system.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 하계학술대회 논문집 A
• /
• pp.480-482
• /
• 2004

This paper presents a study on the NGR problem for harmonic in distribution systems. Overheating of NGR (Neutral Ground reactor), by neutral current in distribution system, is important cause of transformer breakdown of substation. Countermeasures about zero-sequence component harmonic in neutral line are required.

• 산업경영시스템학회지
• /
• 제19권37호
• /
• pp.263-269
• /
• 1996

The cost and customer service level of a logistics system depend primarily on the system design of the physical supply system and physical distribution system. The study presents the mathematical model and a huristic solution method of a combined location - vehicle routing problem(LVRP). In LVRP the objective is to determine the number and location of the distribution centers, the allocation of customers to distribution centers, and the vehicle delivery routes, so as to minimize the logistics total cost and satisfy the customer.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2001년도 춘계학술대회 논문집
• /
• pp.72-79
• /
• 2001

In this paper we present a nonlinear programming model for the solution of the train seat capacity distribution problem (TSCDP) with a numerical example. The TSCDP model finds the optimal capacity distribution methods which minimize the sum of the differences between the demands and the seat capacities. Also the TSCDP provides the information on the degree of the discrepancy between the demand and the seat capacities. One can use the TSCDP model as a tool for planning train seat capacity planning.