http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
전략적 제휴를 통한 신약 개발의 기술가치평가 분석 : 공동연구개발을 중심으로
김휘수,연주한,김천규 한국기술혁신학회 2023 기술혁신학회지 Vol.26 No.5
In the pharmaceutical industry, which invests significant time and resources into generating high value-added products, strategic alliances between companies have become an essential and inevitable choice in order to gain a competitive advantage. While many studies have shown that the outcomes of pharmaceutical alliances vary depending on their type and target, there have been no previous research cases analyzing strategic alliances from a valuation perspective. Therefore, this study aimed to analyze and verify the efficiency of joint research and development in new drug development by using r-NPV (risk-adjusted net present value), a valuation method commonly used in pharmaceutical and biotech industries, to compare the value of a drug developed by Company A through independent research and development versus cooperative research and development. The results showed that the technology value increased 2.3 times as the cumulative new drug approval rate increased in joint research and development, indicating that this could serve as a basis for future strategic alliance activities in the pharmaceutical industry.
임무에 따른 하박 교체형 고 가반하중 양팔로봇의 설계: 구난 및 물체 핸들링
김휘수,박동일,최태용,도현민,김두형,경진호,박찬훈,Kim, Hwisu,Park, Dongil,Choi, Taeyong,Do, Hyunmin,Kim, Doohyeong,Kyung, Jinho,Park, Chanhun 한국로봇학회 2017 로봇학회 논문지 Vol.12 No.4
Robot arms are being increasingly used in various fields with special attention given to unmanned systems. In this research, we developed a high payload dual-arm robot, in which the forearm module is replaceable to meet the assigned task, such as object handling or lifting humans in a rescue operation. With each forearm module specialized for an assigned task (e.g. safety for rescue and redundant joints for object handling task), the robot can conduct various tasks more effectively than could be done previously. In this paper, the design of the high payload dual-arm robot with replaceable forearm function is described in detail. Two forearms are developed here. Each of forearm has quite a different goal. One of the forearms is specialized for human rescue in human familiar flat aspect and compliance parts. Other is for general heavy objects, more than 30 kg, handling with high degree of freedom more than 7.
가변 가반하중을 포함하는 로봇의 필요토크 최소화를 위한 스프링 기반 가변 자중보상 메커니즘의 설계
김휘수,배명수,박찬훈,김두형,경진호,도현민,박동일,최태용,송성혁 제어·로봇·시스템학회 2020 제어·로봇·시스템학회 논문지 Vol.26 No.12
. Recently, with the increasing requirement of human-robot cooperation, robot safety has become an important issue in the robotics field. Counterbalance mechanisms were proposed as a solution to address this problem as it can ensure intrinsic safety and minimize the required torque of the actuators. However, previous counterbalance mechanisms were designed to compensate for the gravitational torque generated due to the weight of robots. Therefore, actuators should be selected to provide sufficient torque to support the payload. In this study, we propose a variable counterbalance mechanism for counterbalance mechanism improvement. The proposed mechanism can adjust the counterbalancing torque according to the weight of the payload attached at the end effector of the robot arm, such as a tool, gripper, or grasped object. Therefore, the torque required to support not only the self-weight of the robot arm but also any additional external load can be compensated effectively using the proposed variable counterbalance mechanism. Moreover, a 1 degree-of-freedom variable counterbalance arm was constructed to verify the effectiveness of the proposed mechanism. Various simulations and experimental results showed that the proposed variable counterbalance mechanism can effectively decrease the torque required to support the robot mass and payload. Hence, low-capacity actuators can be used for designing robots with desired specifications (e.g., maximum payload and repeatability) and safety.
다양한 CNN 가속기에서 아키텍처에 따른 면적, 에너지, 성능 분석
김휘수(Hweesoo Kim),최재완(Jaewan Choi),이선정(Sunjung Lee),안정호(Jung Ho Ahn) 한국정보과학회 2021 정보과학회 컴퓨팅의 실제 논문지 Vol.27 No.1
Convolution Neural Network 가속기는 AI시대에 중요한 요소 중 하나로 떠오르게 되었다. CNN 가속기의 내부 구성을 몇 가지로 나눈다면 연산을 위한 Multiplier-Accumulator (MAC unit), 데이터 저장을 위한 SRAM, 데이터 이동을 위한 메모리 인터페이스 그리고 제어 로직으로 구분할 수 있다. 다양한 CNN 가속기들의 경우, 각기 다른 공정과 동작 주파수를 기준으로 제안되었으며, 또한 아키텍처 형태에 따라 내부 MAC unit의 수와 SRAM의 크기가 매우 큰 차이를 갖는 형태로 구성되어있다. 각 가속기들의 기본 사양으로 면적, 에너지, 성능을 비교하였을 때는 공정이나 동작 주파수 등 여러 조건들에 의해서 아키텍처에 따른 정량적인 비교가 용이하지 않게 된다. 따라서, 본 논문에서는 다양한 CNN 가속기에서 여러 조건들을 동일하게 재구성하였을 때, ResNet-50 추론 동작 시에 요구되는 면적, 에너지, 성능을 비교하여 아키텍처의 특징과 경향성을 분석하였다. The Convolution Neural Network accelerator has emerged as an important element in the AI era. The primary components of a CNN accelerator include the Multiplier-Accumulator (MAC unit) for calculation, SRAM for data storage, memory interface for data movement, and control logic. Different CNN accelerators have been designed based on different assumptions regarding the process technologies and operating frequencies. In addition, the number of internal MAC units and the size of SRAM vary substantially between different types of architectures. These factors make it difficult to design a fair comparison of the area, energy, and performance of different CNN accelerators. In this paper, we attempt to compare the area, energy, and performance of different CNN accelerator architectures by constructing them all with the same fabrication process and operating frequency while making inferences using the ResNet-50 network.
어텐션 기반 협업형 소스측 분산 서비스 거부 공격 탐지
김휘수 ( Hwisoo Kim ),정송헌 ( Songheon Jeong ),김경백 ( Kyungbaek Kim ) 한국정보처리학회 2024 정보처리학회논문지. 소프트웨어 및 데이터 공학 Vol.13 No.4
분산 서비스 거부 공격(DDoS Attack, Distributed Denial of Service Attack) 수법의 진화는 탐지 과정에서의 어려움을 가중시켰다. 기존 피해자측 탐지 방식의 한계로 인해 발생하는 문제를 극복하기 위한 솔루션 중 하나가 소스측 탐지 기법이었다. 그러나 네트워크 트래픽의 불규칙성으로 인한 성능 저하 문제가 존재하였다. 이 문제를 해결하기 위해 인공지능을 기반으로 한 여러 노드 간의 협업 네트워크를 활용하여 공격을 탐지하려는 연구가 진행되었다. 기존의 방법들은 특히 높은 버스트(Burstness)와 지터(Jitter)의 비선형적 트래픽 환경에서 한계를 보였다. 이러한 문제점을 극복하기 위해 본 논문에서는 어텐션(Attention) 메커니즘을 도입한 협업형 소스측 DDoS 공격 탐지 기법을 제시한다. 제안하는 방식은 여러 소스에서의 탐지 결과를 집계하여 각 지역에 가중치를 할당하며, 이를 통해 전반적인 공격 및 특정 소수 지역에서의 공격을 효과적으로 탐지할 수 있다. 특히, 비선형적인 트래픽 데이터셋에서 약 6% 수치의 낮은 가양성(False Positive)과 최대 4.3% 수치가 향상된 높은 탐지율을 보이며, 기존 비선형적 트래픽 환경에서 한계를 보였던 방법들에 비해 소수 지역의 공격 탐지 문제에 대한 개선도 확인할 수 있다. The evolution of the Distributed Denial of Service Attack(DDoS Attack) method has increased the difficulty in the detection process. One of the solutions to overcome the problems caused by the limitations of the existing victim-side detection method was the source-side detection technique. However, there was a problem of performance degradation due to network traffic irregularities. In order to solve this problem, research has been conducted to detect attacks using a collaborative network between several nodes based on artificial intelligence. Existing methods have shown limitations, especially in nonlinear traffic environments with high Burstness and jitter. To overcome this problem, this paper presents a collaborative source-side DDoS attack detection technique introduced with an attention mechanism. The proposed method aggregates detection results from multiple sources and assigns weights to each region, and through this, it is possible to effectively detect overall attacks and attacks in specific few areas. In particular, it shows a high detection rate with a low false positive of about 6% and a high detection rate of up to 4.3% in a nonlinear traffic dataset, and it can also confirm improvement in attack detection problems in a small number of regions compared to methods that showed limitations in the existing nonlinear traffic environment.
김휘수(Hwi-Su Kim),박정준(Jung-Jun Park),송재복(Jae-Bok Song) 대한기계학회 2009 대한기계학회 춘추학술대회 Vol.2009 No.5
In recent years, collision safety between humans and robots has drawn much attention since cooperation between humans and robots is increasingly needed in various fields. Since both positioning accuracy and collision safety are important, an industrial manipulator should maintain very high stiffness for positioning accuracy in the normal situation, but exhibit very low stiffness when subjected to a collision force greater than the injury tolerance. In order to achieve these requirements of an industrial manipulator, a safety mechanism composed of a linear spring and a double-slider mechanism was proposed in the previous research. To satisfy the operational requirements of an industrial manipulator, the spring stiffness of this safety mechanism was designed by conducting various dynamic analyses. Several experiments on static collisions show high stiffness of the safety mechanism against an external torque less than the pre-determined threshold torque, but the absorption of collision force when the external torque exceeds this threshold torque, thereby guaranteeing positioning accuracy and collision safety.