• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.4A
• /
• pp.301-309
• /
• 2002

In this paper, we propose the effective implementation of various nonlinear combiners using by PS-LFSR: m-parallel memoryless-nonlinear combiner, m-parallel memory-nonlinear combiner, m-parallel nonlinear filter function, and m-parallel clock-controlled function. Finally, we propose m-parallel LILI-128 stream cipher as an example of the parallel implementation, and we determine its cryptographic security and performance.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.4
• /
• pp.859-867
• /
• 2010

In this paper, we propose some parallel structures of the word-based nonlinear combining functions in word-based stream cipher, high-speed versions of general (bit-based) nonlinear combining functions. Especially, we propose the high-speed structures of popular four kinds in word-based nonlinear combiners using by PS-WFSR (Parallel-Shifting or Parallel-Structured Word-based FSR): m-parallel word-based nonlinear combiner without memory, m-parallel word-based nonlinear combiner with memories, m-parallel word-based nonlinear filter function, and m-parallel word-based clock-controlled function. In addition, we propose an implementation example of the m-parallel word-based DRAGON stream cipher, and determine its cryptographic security and performance.

• Proceedings of the Korea Society for Industrial Systems Conference
• /
• 2001.05a
• /
• pp.77-83
• /
• 2001

In recent years, the AES in North America and the NESSIE project in Europe have been in progress. Six proposals have been submitted to the NESSIE project including the LILI-128 by Simpson in Australia in the synchronous stream cipher category. These proposals tend towards a design with parallelism of the algorithms in order to facilitate speed-up. In this paper, we consider the PS-LFSR and propose the effective implementation of various nonlinear combiners: memoryless-nonlinear combiner, memory-nonlinear combiner, nonlinear filter function, and clock-controlled function. Finally, we propose m-parallel SUM-BSG and LILI-l28's parallel implementation as examples, and we determine their securities and performances.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2009.10a
• /
• pp.383-386
• /
• 2009

In this paper, we propose some parallel structures of the word-based nonlinear combine functions in word-based stream cipher, high-speed versions of general (bit-based) nonlinear combine functions. Especially, we propose the high-speed structures of popular three kinds in word-based nonlinear combiners using by PS-WFSR (Parallel-Shifting or Parallel-Structured Word-based FSR): m-parallel word-based nonlinear combiner without memory, m-parallel word-based nonlinear combiner with memories, and m-parallel word-based nonlinear filter function. Finally, we analyze its cryptographic security and performance.

• Communications for Statistical Applications and Methods
• /
• v.3 no.1
• /
• pp.229-234
• /
• 1996

로우버스트 추정에서 자주 사용되는 Huber의 ${\phi}$-함수의 선형 결합을 통해 새로운 재하강 ${\phi}$-함수를 도출한다. 이 함수를 사용하면 적절한 조건하에서 앞의 두 함수를 사용할 때보다 위치 모수(location parameter)에 대한 추정량의 점근분산(asymptotic variance)을 감소시킬 수 있음을 보였다.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2015.08a
• /
• pp.140.2-140.2
• /
• 2015

낮은 압력의 평판형 유도 결합 플라즈마 (Inductively Coupled Plasma, ICP)에서 챔버 높이를 바꾸면서 전자 에너지 확률 함수 (Electron Energy Probability Function, EEPF)를 측정하였다. 측정된 전자 에너지 확률 함수에서 기울기가 평평한 부분이 관찰됐고, 이러한 전자 에너지 분포함수의 평평한 부분은 챔버 높이를 증가함에 따라 높은 전자 에너지로 옮겨졌다. 이러한 현상을 분석하기 위해서 2차원 비충돌 가열 메커니즘이 포함된 유도 결합 플라즈마 모델로부터 전자 에너지 확산 계수와 이론적인 전자 에너지 확률 함수를 구하여 실험 결과와 비교하였다. 이를 통하여, 측정된 전자 에너지 확률 함수의 평평한 부분은 전자 튕김 공진 (electron bounce resonance)에 의한 것임을 알 수 있었다.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.9
• /
• pp.371-380
• /
• 2022

Convolutional neural networks are widely used to manipulate data arranged in a grid, such as images. A general convolutional neural network consists of a convolutional layers and a fully connected layers, and each layer contains a nonlinear activation functions. This paper proposes a combined parametric activation function to improve the performance of convolutional neural networks. The combined parametric activation function is created by adding the parametric activation functions to which parameters that convert the scale and location of the activation function are applied. Various nonlinear intervals can be created according to parameters that convert multiple scales and locations, and parameters can be learned in the direction of minimizing the loss function calculated by the given input data. As a result of testing the performance of the convolutional neural network using the combined parametric activation function on the MNIST, Fashion MNIST, CIFAR10 and CIFAR100 classification problems, it was confirmed that it had better performance than other activation functions.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.1
• /
• pp.1-10
• /
• 2022

Deep neural networks are widely used to solve various problems. In a fully connected neural network, the nonlinear activation function is a function that nonlinearly transforms the input value and outputs it. The nonlinear activation function plays an important role in solving the nonlinear problem, and various nonlinear activation functions have been studied. In this study, we propose a combined parametric activation function that can improve the performance of a fully connected neural network. Combined parametric activation functions can be created by simply adding parametric activation functions. The parametric activation function is a function that can be optimized in the direction of minimizing the loss function by applying a parameter that converts the scale and location of the activation function according to the input data. By combining the parametric activation functions, more diverse nonlinear intervals can be created, and the parameters of the parametric activation functions can be optimized in the direction of minimizing the loss function. The performance of the combined parametric activation function was tested through the MNIST classification problem and the Fashion MNIST classification problem, and as a result, it was confirmed that it has better performance than the existing nonlinear activation function and parametric activation function.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.6B
• /
• pp.840-847
• /
• 2001

본 논문에서는 클럭 동기 회로에 사용되는 다차 함수 형태의 결합 필터를 선형 근사화 하는 알고리즘을 제안하고 이를 하드웨어로 구현한다. 정합 필터와 보간필터에 의한 클럭 동기회로는 수신기를 전 디지털 회로를 구현하기 위해 선호되지만 계산량이 증가하는 단점이 있다. 본 논문에서는 정합 필터의 임펄스 응답을 갖는 결합 보간 필터를 구현하고, base 함수의 적용을 선형 근사화 하여 필터의 계산량을 감소시켰다. 본 논문에서는 선형 근사화된 결합 보간 필터의 동작을 Matlab을 통한 시뮬레이션과 ALTERA Chip으로 테스트하였다.

• Bioinformatics and Biosystems
• /
• v.1 no.2
• /
• pp.103-108
• /
• 2006

Computational prediction of protein-ligand binding has been widely used as a tool to discover lead compounds fur new drugs. Prediction accuracy is determined in part by the scoring function used in docking calculations. Diverse scoring functions are available, and these can be classified into force-field based, empirical, and knowledge-based functions depending upon the basic assumptions made in development. Among these, force-field based functions consider physical interactions the most in detail. However, the force-field based functions have the drawback of not including the entropic effect while considering only the energy contribution such as dispersion or electrostatic forces. In this article, a method to take into account of the entropic effect using the colony energy is suggested when force-field based scoring functions is used by extracting conformational information obtained from the pre-existing docking program. An improved result for decoy discrimination is illustrated when the method is applied to the DOCK scoring function, and this implies that more accurate docking calculation is possible.