• 정보보호학회논문지
• /
• 제20권1호
• /
• pp.123-128
• /
• 2010

공격 시나리오의 그래프 표현은 서버의 취약성 분석 및 공격의 방어를 위한 설계에 필수적인 방법이다. 이를 위해 다양한 요구사항 분석 모델이 이용되고 있으나, 복잡한 시나리오간의 결합을 표현할 수 있는 모델은 제한적이다. 본 논문에서 제안하는 역할 기반 페트리 넷(Role Based Petri Net)은 동시성과 시각적인 장점을 가진 페트리 넷을 역할 기반으로 구성하여 효과적 표현 모델을 제공하고 알려지지 않은 공격에 대한 시나리오를 효율적으로 표현할 수 있다.

• 전자공학회논문지
• /
• 제53권1호
• /
• pp.3-8
• /
• 2016

무선망에서 소모 전력, 전송률, 통신 반경 등 서비스 요구 사항에 따라 상호 다른 망의 발전이 진행되어 왔다. 최근 차세대 무선망 환경에서는 초고속, 초저지연, 저전력 등 서비스 요구 사항이 보다 다양해지고 높은 수준으로 설정되고 있으며, 이를 만족하기 위한 방안으로 이기종간 효과적인 연동에 대한 연구에 관심이 높아지고 있다. 본 논문에서는 이기종 무선망 환경에서 망간 연동을 통해 데이터 서비스가 이루어지는 상황에서 에너지 효율성을 반영하는 스케쥴링 기법을 제안한다. 특히 이기종 무선망 환경에서 사용자 형평성을 고려하면서 데이터 수율을 개선하는 비례균등 스케쥴링 방식을 기반으로 에너지 효율에 관한 요소를 반영하는 문제를 고려하고 에너지 효율을 개선함과 동시에 사용자 형평성, 데이터 수율을 모두 감안하는 에너지 효율적인 비례균등 스케쥴링 방식을 제안한다. 또한 모의실험을 통해 제안된 방식으로 비례균등 달성도를 유지하면서 에너지 효율이 개선됨을 보인다.

• 산업경영시스템학회지
• /
• 제46권1호
• /
• pp.55-67
• /
• 2023

Lightweight face recognition models, as one of the most popular and long-standing topics in the field of computer vision, has achieved vigorous development and has been widely used in many real-world applications due to fewer number of parameters, lower floating-point operations, and smaller model size. However, few surveys reviewed lightweight models and reimplemented these lightweight models by using the same calculating resource and training dataset. In this survey article, we present a comprehensive review about the recent research advances on the end-to-end efficient lightweight face recognition models and reimplement several of the most popular models. To start with, we introduce the overview of face recognition with lightweight models. Then, based on the construction of models, we categorize the lightweight models into: (1) artificially designing lightweight FR models, (2) pruned models to face recognition, (3) efficient automatic neural network architecture design based on neural architecture searching, (4) Knowledge distillation and (5) low-rank decomposition. As an example, we also introduce the SqueezeFaceNet and EfficientFaceNet by pruning SqueezeNet and EfficientNet. Additionally, we reimplement and present a detailed performance comparison of different lightweight models on the nine different test benchmarks. At last, the challenges and future works are provided. There are three main contributions in our survey: firstly, the categorized lightweight models can be conveniently identified so that we can explore new lightweight models for face recognition; secondly, the comprehensive performance comparisons are carried out so that ones can choose models when a state-of-the-art end-to-end face recognition system is deployed on mobile devices; thirdly, the challenges and future trends are stated to inspire our future works.

• 한국CDE학회논문집
• /
• 제10권3호
• /
• pp.168-178
• /
• 2005

The Delaunay triangular net is primarily characterized by a balance of the whole by improving divided triangular patches into a regular triangle, which closely resembles an equiangular triangle. A triangular net occurring in certain, point-clustered, data is unique and can always create the same triangular net. Due to such unique characteristics, Delaunay triangulation is used in various fields., such as shape reconstruction, solid modeling and volume rendering. There are many algorithms available for Delaunay triangulation but, efficient sequential algorithms are rare. When these grids involve a set of points whose distribution are not well proportioned, the execution speed becomes slower than in a well-proportioned grid. In order to make up for this weakness, the ids are divided into sub-grids when the sets are integrated inside the grid. A method for finding a mate in an incremental construction algorithm is to first search the area with a higher possibility of forming a regular triangular net, while the existing method is to find a set of points inside the grid that includes the circumscribed sphere, increasing the radius of the circumscribed sphere to a certain extent. Therefore, due to its more efficient searching performance, it takes a shorer time to form a triangular net than general incremental algorithms.

• 전기전자학회논문지
• /
• 제24권3호
• /
• pp.685-690
• /
• 2020

휴대용 기기나 에지 단말을 위한 CNN인 MobileNet V2를 기반으로 연산량을 크게 줄이면서도 정확도는 증가시킨 효율적인 인공신경망 네트워크 구조를 제안한다. 제안하는 구조는 Bottleneck 층 구조를 유지하면서 확장 계수를 증가시키고 일부 층을 제거하는 등의 변화를 통해 연산량을 절반 이하로 줄였다. 설계한 네트워크는 ImageNet100 데이터셋을 이용하여 분류 정확도와 CPU 및 GPU에서의 연산 시간을 측정하여 그 성능을 검증 하였다. 또한, 현재 딥러닝 가속기로 널리 이용하는 GPU에서 네트워크 구조에 따라 동작 성능이 달라짐도 보였다.

• 수산경영론집
• /
• 제42권3호
• /
• pp.31-39
• /
• 2011

This paper attempts to extend such analysis to the rather more difficult problem of optimal management of transnational fish stocks jointly owned by two countries. Transboundary fish such as Mackerel creates an incentive to harvest fish before a competitor does and leads to over-exploitation. This tendency is especially poignant for transnational stocks since, in the absence of an enforceable, international agreement, there is little or no reason for either government or the fishing industry to promote resource conservation and economic efficiency. In the current paper I examine a game theoretic setting in which cooperative management can provide more benefits than noncooperative management. A dynamic model of Mackerel fishery is combined with Nash's theory of two countries cooperative games. A characteristic function game approach is applied to describe the sharing of the surplus benefits from cooperation and noncooperation. A bioeconomic model was used to compare the economic yield of the optimal strategies for two countries, under joint maximization of net benefits in joint ocean. The results suggest as follows. First, the threat points represent the net benefits for two countries in absence of cooperation. The net benefits to Korea and China in threat points are 2,000 billion won(${\pi}^0_{KO}$) and 1,130 billion won(${\pi}^0_{CH}$). Total benefits are 3,130 billion won. Second, if two countries cooperate one with another, they reach the solution payoffs such as Pareto efficient. The net benefits to Korea and China in Pareto efficient are 2,785 billion won(${\pi}^0_{KO}$) and 1,605 billion won(${\pi}^0_{CH}$) or total benefits of 4,390 billion won : a gain of 1,260 billion won. Third, the different price effects under the two scenarios show that total benefit rise as price increases.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2021년도 춘계학술대회
• /
• pp.374-376
• /
• 2021

임상에서 암 관련 질병의 확진을 위해 영상장비를 이용한 기초 진단 이후 추가적인 방법으로 생체검사 등을 이용한 병리적 검사가 필수적이다. 이러한 생체검사를 진행하기 위해서는 전문지식을 가진 종양학자, 임상병리사 등의 도움과 최소한의 소요시간은 확진을 위해 반드시 필요하다. 최근 들어, 인공지능을 활용한 암세포의 자동분류가 가능한 시스템 구축에 관련된 연구가 활발하게 진행되고 있다. 하지만, 이전 연구들은 한정된 알고리즘을 기반으로 하여 세포의 종류와 정확도에 한계를 보인다. 본 연구에서 심층 학습의 일종인 합성곱 신경망을 통해 총 4가지의 암세포를 식별하는 방법을 제안한다. 세포 배양을 통해 얻은 광학영상을 OpenCV를 사용하여 세포의 위치 식별 및 이미지 분할과 같은 전처리 수행 후, EfficientNet을 통해 학습하였다. 모델은 EfficientNet을 기준으로 다양한 hyper parameter를 사용하고, InceptionV3을 학습하여 성능을 비교분석 하였다. 그 결과 96.8%의 높은 정확도로 세포를 분류하는 결과를 보였으며, 이러한 분석방법은 암의 확진에 도움이 될 것으로 기대한다.

• 한국정밀공학회지
• /
• 제25권3호
• /
• pp.147-156
• /
• 2008

A new approach based on cone prism intersection method combined with sorting algorithm is proposed for the fast and robust signed distance field computation. In the method, the space bounding the geometric model composed of triangular net is divided into multiple smaller cells. For the efficient calculation of distance fields, valid points among the triangular net which will generate minimum distances with current cell are selected by checking the intersection between current cell and cone prism generated at each point. The method is simple to implement and able to achieve an order of magnitude improvement in the computation time as compared to earlier approaches. Further the method is robust in handling the traditional sign problems. The validity of the suggested method was demonstrated by providing numerous examples including Boolean operation, shape deformation and morphing of complex geometric models.

• Journal of Computing Science and Engineering
• /
• 제7권1호
• /
• pp.44-52
• /
• 2013

This paper addresses the problem of resource scheduling in a grid computing environment. One of the main goals of grid computing is to share system resources among geographically dispersed users, and schedule resource requests in an efficient manner. Grid computing resources are distributed, heterogeneous, dynamic, and autonomous, which makes resource scheduling a complex problem. This paper proposes a new approach to resource scheduling in grid computing environments, the hierarchical stochastic Petri net (HSPN). The HSPN optimizes grid resource sharing, by categorizing resource requests in three layers, where each layer has special functions for receiving subtasks from, and delivering data to, the layer above or below. We compare the HSPN performance with the Min-min and Max-min resource scheduling algorithms. Our results show that the HSPN performs better than Max-min, but slightly underperforms Min-min.

• Journal of Multimedia Information System
• /
• 제8권3호
• /
• pp.175-182
• /
• 2021

Oral cancer is ranked second most diagnosed cancer among Indian population and ranked sixth all around the world. Oral cancer is one of the deadliest cancers with high mortality rate and very less 5-year survival rates even after treatment. It becomes necessary to detect oral malignancies as early as possible so that timely treatment may be given to patient and increase the survival chances. In recent years deep learning based frameworks have been proposed by many researchers that can detect malignancies from medical images. In this paper we have proposed a deep learning-based framework which detects oral cancer from histopathology images very efficiently. We have designed our model to split the color channels and extract deep features from these individual channels rather than single combined channel with the help of Efficient NET B3. These features from different channels are fused by using feature fusion module designed as a layer and placed before dense layers of Efficient NET. The experiments were performed on our own dataset collected from hospitals. We also performed experiments of BreakHis, and ICML datasets to evaluate our model. The results produced by our model are very good as compared to previously reported results.