• Journal of Internet Computing and Services
• /
• v.19 no.6
• /
• pp.83-89
• /
• 2018

In workflow process intelligence and systems, workflow process mining and analysis issues are becoming increasingly important. In order to improve the quality of workflow process intelligence, it is essential for an efficient and effective data center storing workflow enactment event logs to be provisioned in carrying out the workflow process mining and analytics. In this paper, we propose a three-dimensional process-aware datacube for organizing workflow enterprise data centers to efficiently as well as effectively store the workflow process enactment event logs in the XES format. As a validation step, we carry out an experimental process mining to show how much perfectly the process-aware datacubes are suitable for discovering workflow process patterns and its analytical knowledge, like enacted proportions and enacted work transferences, from the workflow process enactment event histories. Finally, we confirmed that it is feasible to discover the fundamental control-flow patterns of workflow processes through the implemented workflow process mining system based on the process-aware data cube.

• KIPS Transactions on Computer and Communication Systems
• /
• v.11 no.7
• /
• pp.205-216
• /
• 2022

One of the many tools used for distributed deep learning training is Kubeflow, which runs on Kubernetes, a container orchestration tool. TensorFlow jobs can be managed using the existing operator provided by Kubeflow. However, when considering the distributed deep learning training jobs based on the parameter server architecture, the scheduling policy used by the existing operator does not consider the task affinity of the distributed training job and does not provide the ability to dynamically allocate or release resources. This can lead to long job completion time and low resource utilization rate. Therefore, in this paper we proposes a new operator that efficiently schedules distributed deep learning training jobs to minimize the job completion time and increase resource utilization rate. We implemented the new operator by modifying the existing operator and conducted experiments to evaluate its performance. The experiment results showed that our scheduling policy improved the average job completion time reduction rate of up to 84% and average CPU utilization increase rate of up to 92%.

• Journal of Industrial Technology
• /
• v.15
• /
• pp.223-233
• /
• 1995

본 논문에서는 데이터 분포가 편재된 상황하에서 부하의 불균형과 버켓 오벌플로우 문제를 해결하기 위해 히스토그램 변환 기법을 이용한 데이터 분산 방법과 이를 기초로 한 병렬 결합 알고리즘을 제안한다. 제안된 알고리즘의 성능은 시뮬레이션과 하이퍼큐브형 병렬 컴퓨터 상에서 실험적인 방법에 의하여 분석되었다. 그 결과 제안된 알고리즘이 기본의 해쉬 결합 방법보다 우수함을 보인다.

• The Journal of Engineering Geology
• /
• v.32 no.1
• /
• pp.59-72
• /
• 2022

This study employed a 3-D numerical analysis based on the distinct element method to estimate the strength and deformability of a Cretaceous biotite granitic rock mass at Gijang, Busan, Korea. A workflow was proposed to evaluate the scale effect and the representative elementary volume (REV) of mechanical properties for fractured rock masses. Directional strength and deformability parameters such as block strength, deformation modulus, shear modulus, and bulk modulus were estimated for a discrete fracture network (DFN) in a cubic block the size of the REV. The size of the mechanical REV for fractured rock masses in the study area was determined to be a 15 m cube. The mean block strength and mean deformation modulus of the DFN cube block were found to be 52.8% and 57.7% of the intact rock's strength and Young's modulus, respectively. A constitutive model was derived for the study area that describes the linear-elastic and orthotropic mechanical behavior of the rock mass. The model is expected to help evaluate the stability of tunnels and underground spaces through equivalent continuum analysis.