• Journal of Korea Entertainment Industry Association
• /
• v.14 no.4
• /
• pp.349-354
• /
• 2020

This paper, firstly, propose a new Clustering scheme Based on Conceptual graphs (CBC) that can describe objects freely and can perform clustering efficiently. The conceptual clustering is one of machine learning technique. The similarity among the objects in conceptual clustering are decided on the bases of concept membership, unlike the general clustering scheme which decide the similarity without considering the context or environment of the objects. A new conceptual clustering scheme, CBC, which can perform efficient conceptual clustering by describing various objects freely with conceptual graphs is introduced in this paper.

• International journal of advanced smart convergence
• /
• v.12 no.3
• /
• pp.75-88
• /
• 2023

In a specific domain, experts have different understanding of domain knowledge or different purpose of constructing ontology. These will lead to multiple different ontologies in the domain. This phenomenon is called the ontology heterogeneity. For research fields that require cross-ontology operations such as knowledge fusion and knowledge reasoning, the ontology heterogeneity has caused certain difficulties for research. In this paper, we propose a novel ontology matching model that combines word embedding and a concatenated continuous bag-of-words model. Our goal is to improve word vectors and distinguish the semantic similarity and descriptive associations. Moreover, we make the most of textual and structural information from the ontology and external resources. We represent the ontology as a graph and use the SimRank algorithm to calculate the structural similarity. Our approach employs a similarity queue to achieve one-to-many matching results which provide a wider range of insights for subsequent mining and analysis. This enhances and refines the methodology used in ontology matching.

• Annual Conference of KIPS
• /
• 2022.05a
• /
• pp.557-559
• /
• 2022

최근 딥러닝의 기술발전으로 자연어 처리 분야에서 Q&A, 문장추천, 개체명 인식 등 다양한 연구가 진행 되고 있다. 딥러닝 기반 자연어 처리에서 좋은 성능을 보이는 트랜스포머 기반 BERT 모델의 성능향상에 대한 다양한 연구도 함께 진행되고 있다. 본 논문에서는 토픽모델인 잠재 디리클레 할당을 이용한 토픽별 지식그래프 분류와 입력문장의 토픽을 추론하는 방법으로 K-BERT 모델을 학습한다. 분류된 토픽 지식그래프와 추론된 토픽을 이용해 K-BERT 모델에서 대용량 지식그래프 사용의 효율적 방법을 제안한다.

• Annual Conference of KIPS
• /
• 2022.05a
• /
• pp.583-585
• /
• 2022

기존 군사 분야 영상 판독 시스템은 영상 판독관들의 작업 부담이 크고, 판독관들의 경험과 숙련도에 의존적이다. 이전 연구에서 판독관들의 부담을 줄이고 경험 및 숙련 의존도를 낮추기 위해 문장 추천 시스템을 제안하였다. 하지만 학습에 사용된 데이터의 양이 적고, 학습에 사용되지 않은 장비 혹은 지역 등의 단어가 등장 시 제대로 동작하지 않는 한계점이 있었다. 이를 해결하기 위해 학습 데이터 단계와 디코딩 단계에 지식그래프를 적용하여 문장의 다양성과 확장성을 확보하고, 데이터 부족 문제를 완화하였다. 이 연구는 추후 판독관들의 업무 과부화를 완화하고 업무 효율을 높일 수 있을 것이다.

• Journal of Advanced Information Technology and Convergence
• /
• v.9 no.1
• /
• pp.89-102
• /
• 2019

Collagen can be used in building artificial skin replacements for treatment of burns and towards the reconstruction of bone as well as researching cell behavior and cellular interaction. The strength of collagen in connective tissue rests on the characteristics of collagen fibers. 3D confocal imaging of collagen fibers enables the characterization of their spatial distribution as related to their function. However, the image stacks acquired with confocal laser-scanning microscope does not clearly show the collagen architecture in 3D. Therefore, we developed a new method to reconstruct, visualize and characterize collagen fibers from fluorescence confocal images. First, we exploit the tensor voting framework to extract sparse reliable information about collagen structure in a 3D image and therefore denoise and filter the acquired image stack. We then propose to segment the collagen fibers by defining an energy term based on the Hessian matrix. This energy term is minimized by a min cut-max flow algorithm that allows adaptive regularization. We demonstrate the efficacy of our methods by visualizing reconstructed collagen from specific 3D image stack.

• Journal of Advanced Information Technology and Convergence
• /
• v.9 no.1
• /
• pp.103-113
• /
• 2019

Motivation: Polymerized actin-based cytoskeletal structures are crucial in shape, dynamics, and resilience of a cell. For example, dynamical actin-containing ruffles are located at leading edges of cells and have a significant impact on cell motility. Other filamentous actin (F-actin) bundles, called stress fibers, are essential in cell attachment and detachment. For this reason, their mechanistic understanding provides crucial information to solve practical problems related to cell interactions with materials in tissue engineering. Detecting and counting actin-based structures in a cellular ensemble is a fundamental first step. In this research, we suggest a new method to characterize F-actin wrapping fibers from confocal fluorescence image stacks. As fluorescently labeled F-actin often envelope the fibers, we first propose to segment these fibers by diminishing an energy based on maximum flow and minimum cut algorithm. The actual actin is detected through the use of bilateral filtering followed by a thresholding step. Later, concave actin bundles are detected through a graph-based procedure that actually determines if the considered actin filament is enclosing the fiber.

• Annual Conference of KIPS
• /
• 2022.11a
• /
• pp.266-268
• /
• 2022

최근 블록체인 기술이 부상하면서 이를 이용한 암호화폐가 범죄의 대상이 되고 있다. 특히 피싱 스캠은 이더리움 사이버 범죄의 과반수 이상을 차지하며 주요 보안 위협원으로 여겨지고 있다. 따라서 효과적인 피싱 스캠 탐지 방법이 시급하다. 그러나 전체 노드에서 라벨링된 피싱 주소의 부족으로 인한 데이터 불균형으로 인하여 지도학습에 충분한 데이터 제공이 어려운 상황이다. 이를 해결하기 위해 본 논문에서는 이더리움 트랜잭션 네트워크를 고려한 효율적인 네트워크 임베딩 기법인 trans2vec 과 준지도 학습 모델 tri-training 을 함께 사용하여 라벨링된 데이터뿐만 아니라 라벨링되지 않은 데이터도 최대한 활용하는 피싱 스캠 탐지 방법을 제안한다.

• Journal of information and communication convergence engineering
• /
• v.22 no.3
• /
• pp.215-220
• /
• 2024

This paper presents an efficient method for expanding interconnections in scenarios involving the reconstruction of interconnections across arbitrarily divided planes. Conventionally, such situations necessitate rebuilding interconnections based on all targets, ensuring minimal cost but incurring substantial time expenditure. In this paper, we present a tinkered tree algorithm designed to efficiently expand interconnections within a Euclidean plane divided into m randomly generated regions. The primary objective of this algorithm is to construct an optimal tree by utilizing the minimum spanning tree (MST) of each region, resulting in swift interconnection expansion. Interconnection construction is applied in various design fields. Notably, in the context of ad hoc networks, which lack a fixed-wired infrastructure and communicate solely with mobile hosts, the heuristic proposed in this paper is anticipated to significantly reduce costs while establishing rapid interconnections in scenarios involving expanded connection targets.

• Annual Conference of KIPS
• /
• 2024.10a
• /
• pp.1107-1108
• /
• 2024

4차 산업혁명 시대에 AI 기술의 발전과 함께, 고속 데이터 전송을 위한 6G의 필요성이 대두되고 있으며, 이는 다양한 산업 분야에 큰 영향을 미칠 것으로 기대된다. 그러나 현재의 채널 디코딩 방식인 LDPC 및 BCH 코드 알고리즘은 복잡한 연산으로 인해 실시간 통신에 지연을 초래할 수 있다. GNN은 노드 간의 복잡한 관계를 효과적으로 학습할 수 있어, 통신 채널 특성을 이해하고 예측하는데 유리하다. 본 연구에서는 6G 통신 기술에 접목하기 위해, 기존 디코딩 방식보다 처리속도가 빠르고 비트 오류율이 낮은 그래프 신경망 기반 채널 디코딩 모델 개발을 목표로 한다.

• Journal of KIISE:Computer Systems and Theory
• /
• v.27 no.2
• /
• pp.135-148
• /
• 2000

In this paper, we present a linear scheduling method for homogeneous multiprocessor systems using genetic algorithms. In general, genetic algorithms randomly generate initial strings, which leads to long operation time and slow convergence due to an inappropriate initialization. The proposed algorithm considers communication costs among processors and generates initial strings such that successive nodes are grouped into the same cluster. In the crossover and mutation operations, the algorithm maintains linearity in scheduling by associating a node with its immediate successor or predecessor. Linear scheduling can fully utilize the inherent parallelism of a given program and has been proven to be superior to nonlinear scheduling on a coarse grain DAG (directed acyclic graph). This paper emphasizes the usability of the genetic algorithm for real-time applications. Simulation results show that the proposed algorithm rapidly converges within 50 generations in most DAGs.