http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Phase Separation in Artificial Lipid Membrane Composed of DPPC, DOPC, and Cholesterol
이현로,이요한,최시영 한국공업화학회 2018 한국공업화학회 연구논문 초록집 Vol.2018 No.0
Raft-based heterogeneity is an important feature of cell membrane because it relates to fundamental biological processes such as signal transduction, membrane trafficking, and endocytosis. To understand the specific role of the lipid raft in the cell membrane, the phase separation phenomenon in the membrane should be scrutinized depending on membrane tension, lipid composition, charge density, temperature and so on. In this work, we developed a new and simple method to make freestanding and planar lipid bilayer model for direct visualization and analysis of phase behavior. Interestingly, compared to other lipid bilayer models such as giant unilamellar vesicles and black lipid membrane, we found that our model had different phase behaviors including domain size and shape, which were significantly affected by geometric shape, interfacial tension, lipid composition, and temperature. Finally, we developed its thermodynamic model using Landau theory on the basis of experimental results.
준지도 학습 모델을 사용한 차량 내부 네트워크에서의 이상 징후 탐지
이현로(Hyunro Lee),홍성우(Seongwoo Hong),이승열(Seungyeol Lee),하재철(Jaecheol Ha) 한국산학기술학회 2023 한국산학기술학회논문지 Vol.24 No.2
현재 자동차 내부의 전자 제어 장치(Electronic Control Unit, ECU)간의 통신을 위해 CAN(Controller Area Network) 프로토콜을 많이 사용하고 있다. 하지만 CAN 프로토콜은 메시지 암호화 및 발신자 인증과 같은 보안 기능을 가지고 있지 않아 인가되지 않은 데이터 주입이나 서비스 거부 공격(Denial of Service, DoS) 등과 같은 사이버 보안위협에 취약하다. 따라서 최근에는 자동차의 CAN 네트워크를 보호하기 위한 인공 지능 기반의 침입 탐지 시스템(Intrusion Detection System, IDS)에 대한 연구가 활발하게 진행되고 있다. 본 논문에서는 먼저 CAN 버스의 데이터 트래픽에 대한 메시지 주입 공격을 탐지할 수 있는 지도 학습(supervised learning)을 사용하는 딥러닝 모델인 DCNN(Deep Convolutional Neural network) 기반의 이상 탐지 모델을 구현하였다. 또한, 지도 학습 모델은 학습용 데이터 셋이 많아야 한다는 한계점을 지적하고 이를 보완하기 위해 준지도 학습(semi-supervised learning)을 사용한 딥러닝 모델인 GAN(Generative Adversarial Network) 기반의 이상 탐지 모델을 제안한다. 제안하는 준지도 학습 기반의 이상 탐지 모델은 기존 지도 학습 모델에서 약 20만 개의 데이터로 학습하던 것을 1,000개의 데이터만으로도 서비스 거부 공격과 스푸핑 공격을 99%이상 탐지할 수 있어 효율적인 차량용 이상 징후 탐지 시스템으로 사용할 수 있다. The CAN (Controller Area Network) protocol is used widely for communication between ECUs (Electronic Control Units) in a vehicle network. On the other hand, the CAN protocol is vulnerable to cyber security threats, such as unauthorized data injection and DoS (Denial of Service) attacks, because it does not have security functions, such as message encryption and sender authentication. Therefore, research on an artificial intelligence-based IDS (Intrusion Detection System) for protecting the CAN network has been actively conducted. This paper reports an anomaly detection model based on a DCNN (Deep Convolutional Neural Network), a deep learning model using supervised learning that can detect message injection attacks on data traffic on CAN buses. The supervised learning model requires a large number of training data sets. This paper proposes an anomaly detection model based on GAN (Generative Adversarial Network), a deep learning model using semi-supervised learning, to compensate for this advantage. The proposed anomaly detection model based on semi-supervised learning can be used as an efficient vehicle anomaly detection system because it can detect more than 99% of denial-of-service and spoofing attacks with only 1,000 data instead of learning with about 200,000 data in the existing supervised learning model.
이요한,이현로,김규한,최시영 한국공업화학회 2018 한국공업화학회 연구논문 초록집 Vol.2018 No.0
Cell membrane permeability is one of the crucial parameters in the pharmaceutical industry when designing a new drug. Thus, various in vitro assays have been developed to estimate cell membrane permeability of the drug. Instead of a real cell membrane, many assays use an artificial membrane whose microscopic structures are usually unknown, requiring a more accurate and versatile technique. In this research, we developed a new platform to measure the permeability of drug molecules across a planar phospholipid bilayer with a well-defined area and structure. The entire system was constructed within a conventional UV spectrometer cell, and the transport of drug molecules across the bilayer was recorded by UV absorbance in real time. We obtained a much higher permeability compared to the other assays, and this is related to the thickness of the lipid bilayer. Also, we could observe the real-time permeability changes upon the addition of a membrane-disrupting surfactant.
박수진,이현로,최시영 한국공업화학회 2018 한국공업화학회 연구논문 초록집 Vol.2018 No.0
We visualized the albumin adsorption from the top to the interface with phospholipid monolayer, to understand their dynamics and the effects of the serum albumins on the phospholipids. It is known that the serum albumin in the blood inhibits the lung surfactant and leads to respiratory failure, but it was never investigated before that what happen if albumin in blood flow directly to the alveoli surface through airway, not from the aqueous alveolar fluid. When the droplet of the serum albumin coalesces with phospholipid monolayer, the serum albumin leaves a ‘scar’ at the interface. This scar-like layer has radial fingering patterns, and it is very stable. The structures are not completely disappeared for more than 3 hours even above 20 mN/m, the equilibrium surface pressure of the albumin. The stable albumin layers are observed when the albumin concentration is above 0.05 mg/ml, and it is very critical value to patients considering that ~40 mg/ml of albumin is in the blood.
Formation of stable adhesive water-in-oil emulsions using a phospholipid and cosurfactants
김현준,김규한,이현로,조형찬,정대웅,류지흔,권대갑,최시영 한국공업화학회 2017 Journal of Industrial and Engineering Chemistry Vol.55 No.-
The use of adhesive water-in-oil (W/O) emulsions covered with phospholipids have been limited due to their poor stability. We suggest a new and simple method to create adhesive W/O emulsions stabilized by 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) with two different cosurfactants (docosahexaenoic acid (DHA) and sorbitan oleate (SPAN 80)). Although an adhesive W/O emulsion with DOPC is typically unstable because of its molecular structure, we demonstrate that the addition of cosurfactants whose molecular shapes could be complementary to that of DOPC far better stabilizes W/O emulsions and indeed leads to the production of adhesive emulsions by the formation of a bilayer between two monolayers of each droplet surface.