• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제18권3호
• /
• pp.670-684
• /
• 2024

3D Cross-Modal Retrieval (3DCMR) is a task that retrieves 3D objects regardless of modalities, such as images, meshes, and point clouds. One of the most prominent methods used for 3DCMR is the Cross-Modal Center Loss Function (CLF) which applies the conventional center loss strategy for 3D cross-modal search and retrieval. Since CLF is based on center loss, the center features in CLF are also susceptible to subtle changes in hyperparameters and external inferences. For instance, performance degradation is observed when the batch size is too small. Furthermore, the Mean Squared Error (MSE) used in CLF is unable to adapt to changes in batch size and is vulnerable to data variations that occur during actual inference due to the use of simple Euclidean distance between multi-modal features. To address the problems that arise from small batch training, we propose a Noisy Center Loss (NCL) method to estimate the optimal center features. In addition, we apply the simple Siamese representation learning method (SimSiam) during optimal center feature estimation to compare projected features, making the proposed method robust to changes in batch size and variations in data. As a result, the proposed approach demonstrates improved performance in ModelNet40 dataset compared to the conventional methods.

• 대한수학회논문집
• /
• 제32권2호
• /
• pp.425-433
• /
• 2017

This paper is concerned with the analysis of the busy period distribution in a batch arrival $M^X/G/1$ retrial queue. The expression for the Laplace-Stieltjes transform of the length of the busy period is well known, but from this expression we cannot compute the moments of the length of the busy period by direct differentiation. This paper provides a direct method of calculation for the first and second moments of the length of the busy period.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1991년도 한국자동제어학술회의논문집(국제학술편); KOEX, Seoul; 22-24 Oct. 1991
• /
• pp.1718-1723
• /
• 1991

The determination of a production sequence is an important problem in a batch process operation. In this paper a new algorithm for a near optimal production sequence of N product in an M unit serial multiproduct batch process is proposed. The basic principle is the same as that of Johnson's algorithm for two-unit UIS system. Test results on a number of selected examples exhibit the superiority over previously reported results. In addition, a tabulation technique is presented to calculate the makespan of a given sequence of production for all processing units under UIS mode.

• 한국정보기술응용학회:학술대회논문집
• /
• 한국정보기술응용학회 2005년도 6th 2005 International Conference on Computers, Communications and System
• /
• pp.271-275
• /
• 2005

Queueing models are good models for fragments of communication systems and networks, so their investigation is interesting for theory and applications. Theses queues may play an important role for the validation of different decomposition algorithms designed for investigating more general queueing networks. So, in this paper we illustrate that the batch size distribution affects the loss probability, which is the main performance measure of a finite buffer queues.

• 대한수학회보
• /
• 제47권6호
• /
• pp.1251-1258
• /
• 2010

For an M/G/1 processor-sharing queue with batch arrivals, Avrachenkov et al. [1] conjectured that the conditional mean sojourn time is concave. However, Kim and Kim [5] showed that this conjecture is not true in general. In this paper, we show that this conjecture is true if the service times have a hyperexponential distribution.

• 한국경영과학회:학술대회논문집
• /
• 한국경영과학회 2006년도 추계학술대회
• /
• pp.281-283
• /
• 2006

In this paper, we consider the problem of batch processing of orders, where either a single order or a pair of orders which satisfies specific conditions may be grouped in the same batch. The objective of the problem is to minimize the number of batches formed to accommodate all orders. We prose an approach based on a Known algorithm proven to be optimal for special class of problems with tree structure and show the approach to have the worst case ratio of $2-{\frac{2}{n}}$

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1993년도 한국자동제어학술회의논문집(국내학술편); Seoul National University, Seoul; 20-22 Oct. 1993
• /
• pp.749-753
• /
• 1993

It is very difficult to control batch reactor with conventional linear controller due to its severe nonlinearity. To control the nonlinearity of batch reactor, we applied with relay feedback method and SOAS. The SOAS can be designed to work quite well, but it requires engineering effect and some knowledge about the process in order to get a satisfactory performance of the closed loop system For the applications to more reliable, further studies on robustness in various situations and process noises and would be required.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2001년도 하계학술대회 논문집 C
• /
• pp.1798-1800
• /
• 2001

As a batch Type device, this machine is contrive to seal glass plates for plasma BLU (Back Light Unit) by indirect heating from electric heaters. In order to maintain the heating / cooling Chambers clean, this machine uses a muffle formation. The components of the machine are listed bellow.

• 대한산업공학회지
• /
• 제22권1호
• /
• pp.95-104
• /
• 1996

We consider a two-phase queueing system with generalized vacation. Poisson arrivals receive a batch type service in the first phase and individual services in the second phase. The server takes generalized vacation when the system becomes empty. Generalized vacation includes single vacation, multiple vacation, and other types. We consider both gated batch service and exhaustive batch service. This is an extension of the model presented by Selvam and Sivasankaran [6].

• 한국산업정보학회논문지
• /
• 제11권1호
• /
• pp.1-6
• /
• 2006

Queueing models are good models for fragments of communication systems and networks, so their investigation is interesting for theory and applications. Theses queues may play an important role for the validation of different decomposition algorithms designed for investigating more general queueing networks. So, in this paper we illustrate that the batch size distribution affects the loss probability, which is the main performance measure of a finite buffer queues.