• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2001년도 추계학술대회 학술발표 논문집
• /
• pp.255-258
• /
• 2001

This paper is implementation of cellular automata neural network system which is a living creatures' brain using evolving hardware concept. Cellular automata neural network system is based on the development and the evolution, in other words, it is modeled on the ontogeny and phylogeny of natural living things. The proposed system developes each cell's state in neural network by CA. And it regards code of CA rule as individual of genetic algorithm, and evolved by genetic algorithm. In this paper we implement this system using evolving hardware concept Evolving hardware is reconfigurable hardware whose configuration is under the control of an evolutionary algorithm. We design genetic algorithm process for evolutionary algorithm and cells in cellular automata neural network for the construction of reconfigurable system. The effectiveness of the proposed system is verified by applying it to time-series prediction.

• 한국지능시스템학회논문지
• /
• 제11권5호
• /
• pp.426-432
• /
• 2001

본 논문은 진화형 하드웨어를 이용하여 생물의 정보처리 시스템인 셀룰라 오토마타 신경망의 구현에 관한 연구이다. 셀룰라 오토마타 신경망은 진화 및 발생을 기반으로 한 신경망 모델이다. 진화는 다양성을 주요 근원을 제공하는 돌연변이 및 재 조합 비율에 의하여 비결정론이며, 발생은 결정론 적이며 지역적인 무리현상을 따른다. 셀룰라 오토마타 신경망은 셀룰라 오토마타에 의해 신경망 내부의 각 셀의 상태를 발생시키고, 초기 셀을 유전자 알고리즘의 개체로 간주하여 초기 셀이 진화 알고리즘을 통해 진화함으로써 신경망이 진화하는 시스템이다. 본 논문은 이 시스템을 진화형 하드웨어 이용하여 하드웨어로 구현하였다. 진화형 하드웨어는 진화 알고리즘과 재구성하드웨어의 결합체이다. 즉, 재구성 하드웨어의 구성에 필요한 bit를 유전자 알고리즘의 개체로 간주한 것이다. 진화 알고리즘을 수행하기 위해 유전자 알고리즘 프로세서를 설계하였으며, 셀룰라 오토마타 신경망이 유전자 알고리즘의 개체와 셀룰라 오토마타 룰에 의해 자동적으로 신경망을 생성하기 위해 신경망을 이루는 셀들로 설계하였다. 제안된 시스템의 효율성을 검증하기 위해 Exclusive-OR 문제에 적용하였다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1997년도 한국자동제어학술회의논문집; 한국전력공사 서울연수원; 17-18 Oct. 1997
• /
• pp.1334-1337
• /
• 1997

A spacecraft attitude control ground hardware simulator development is discussed in the paper. The simulator is called KT/KARI HILSSAT(Hardware-In-the Loop Simulator Single Axis Testbed), and the main structure consists of a single axis bearing and a satellite main body model on the bearing. The single axis tabel as ans experimental hardware simulator that evaluates performance and applicability of a satellite before evolving and/or confirming a mew or and old control logic used in the KOREASAT is developed. Attitude control of spaceraft by using reaction wheel is performed.

• 한국지능시스템학회논문지
• /
• 제11권8호
• /
• pp.746-753
• /
• 2001

In this paper, we review the basic concept of Evolvable Hardware first. And we examine genetic algorithm processor and hardware reconfiguration method and implementation. By considering complexity and performance of hardware at the same time, we design genetic algorithm processor using modularization and parallel processing method. And we design frame that has connection structure and logic block on FPGA, and embody reconfigurable hardware that do so that this frame may be reconstructed by RAM. Also we implemented ECANS that information processing system such as living creatures'brain using this hardware reconfiguration method. And we apply ECANS which is implemented using the concept of Evolvable Hardware to time-series prediction problem in order to verify the effectiveness.

• 한국정보과학회논문지:소프트웨어및응용
• /
• 제32권1호
• /
• pp.62-73
• /
• 2005

진화 하드웨어(evolvable hardware)는 재구성 가능한 디지털 회로에 진화연산이 적용되어 실시간으로 환경에 적응함으로써 필요한 기능을 자동적으로 구현하는 기술이다. 이는 하드웨어 회로의 자동설계 가능성을 열어 주었지만, 아직 복잡한 회로를 얻기에는 한계가 있다. 본 논문에서는 진화 하드웨어의 적합도 공간을 분석하여, 다양한 개체가 동시에 진화되는 종분화 기법을 제안하고 그 효율성을 실험적으로 보인다. 또한 종분화 기법으로 얻은 다양한 회로를 분석하여 유용한 부가 기능이 창출될 수 있음을 보인다.

• Interaction and multiscale mechanics
• /
• 제6권2호
• /
• pp.197-210
• /
• 2013

A fundamental trend of processor architecture evolving towards exaflops is fast increasing floating point performance (so-called "free" flops) accompanied by much slowly increasing memory and network bandwidth. In order to fully enjoy the "free" flops, a numerical algorithm of PDEs should request more flops per byte or increase arithmetic intensity. A meshfree/GFEM approximation can be the class of the algorithm. It is shown in a GFEM without extra dof that the kind of approximation takes advantages of the high performance of manycore GPUs by a high accuracy of approximation; the "expensive" method is found to be reversely hardware-efficient on the emerging architecture of manycore.

• 전자통신동향분석
• /
• 제38권5호
• /
• pp.12-22
• /
• 2023

The development of neural networks is evolving towards the adoption of transformer structures with attention modules. Hence, active research focused on extending the concept of lightweight neural network algorithms and hardware acceleration is being conducted for the transition from conventional convolutional neural networks to transformer-based networks. We present a survey of state-of-the-art research on lightweight neural network algorithms and hardware architectures to reduce memory usage and accelerate both inference and training. To describe the corresponding trends, we review recent studies on token pruning, quantization, and architecture tuning for the vision transformer. In addition, we present a hardware architecture that incorporates lightweight algorithms into artificial intelligence processors to accelerate processing.

• 한국지능시스템학회논문지
• /
• 제12권5호
• /
• pp.462-466
• /
• 2002

GA(Genetic Algorithm)는 자연계 진화를 모방한 계산 알고리즘으로서 단순하고 응용이 쉽기 때문에 여러 분야에 전역적 최적해 탐색에 많이 사용되고 있다. 최근에는 하드웨어를 구성하는 방법의 하나로서 사용되어 진화하드웨어라는 분야를 탄생시켰다. 이와 함께 GA의 연산자체를 하드웨어로 구현하는 GA processor(GAP)의 필요성도 증가하고 있다. 특히 진화하드웨어를 소프트웨어에서 진화시키는 것이 아닌 GAP에 의해 진화시키는 것은 독립된 구조의 진정한 EHW 설계에 필수적이 될 것이다. 본 논문에서는 진화하드웨어의 빠른 재구성을 위한 유전자 알고리즘 프로세서를 설계한다.

• 한국컴퓨터정보학회논문지
• /
• 제22권9호
• /
• pp.25-32
• /
• 2017

In this paper, we propose the softwareization of broadcasting system. Recently, the topic of industry is the fourth industrial revolution. The fourth industrial revolution is evolving from physical to virtualization. The Industrial Revolution is based on IT technology. Artificial Intelligence (AI), Big Data, and the Internet of Things, which are famous for Alpha Go, are based on software. Among IT, software is the main driver of industrial terrain change. The systemization of software on the basis of cloud environment is proceeding rapidly. System development through softwarization can reduce time to market lead time, hardware cost reduction and manual operation compared to existing hardware system. By developing and implementing broadcasting system such as IPTV based on cloud, lead time for opening service compared to existing hardware system can be shortened by more than 90% and investment cost can be saved by about 40%. In addition, the area of the system can be reduced by 50%. In addition, efficiency can be improved between infrastructures, shortening of trouble handling and ease of maintenance. Finally, we can improve customer experience through rapid service opening.

• ETRI Journal
• /
• 제42권6호
• /
• pp.965-975
• /
• 2020

This paper proposes a new framework for the development and deployment of honeypots for evolving malware threats. As new technological concepts appear and evolve, attack surfaces are exploited. Internet of things significantly increases the attack surface available to malware developers. Previously independent devices are becoming accessible through new hardware and software attack vectors, and the existing taxonomies governing the development and deployment of honeypots are inadequate for evolving malicious programs and their variants. Malware-propagation and compromise methods are highly automated and repetitious. These automated and repetitive characteristics can be exploited by using embedded reinforcement learning within a honeypot. A honeypot for automated and repetitive malware (HARM) can be adaptive so that the best responses may be learnt during its interaction with attack sequences. HARM deployments can be agile through periodic policy evaluation to optimize redeployment. The necessary enhancements for adaptive, agile honeypots require a new development and deployment framework.