• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2003년도 ISIS 2003
• /
• pp.66-69
• /
• 2003

This paper considers a reinforcement learning(RL) which deals with real environments. Most reinforcement learning studies have been made by simulations because real-environment learning requires large computational cost and much time. Furthermore, it is more difficult to acquire many rewards efficiently in real environments than in virtual ones. The most important requirement to make real-environment learning successful is the appropriate construction of the state space. In this paper, to begin with, I show the basic overview of the reinforcement learning under real environments. Next, 1 introduce a state-space construction method under real environmental which is State Partition Method. Finally I apply this method to a robot navigation problem and compare it with conventional methods.

• Natural Product Sciences
• /
• 제15권3호
• /
• pp.144-150
• /
• 2009

Gradient centrifugal partition chromatography (GCPC) method was developed and applied to isolate 3,5-dimethoxyphenanthrene-2,7-diol (DMP) and batatasin-I (BA-I) from the dichloromethane soluble extract of Dioscorea opposita. In this method, the lower phase of n-hexane-methanol-water system (HMW, 10 : 9 : 1, v/v) was used as a mobile phase A (MpA) and water was used as a mobile phase B (MpB). This gradient CPC method is comparable to that of reversed-phase HPLC method in that the stationary upper-phase of HMW (10 : 9 : 1 v/v) works as if it were reversed-phase silica gel due to its hydrophobic property, while the lower phase (MpA) and water (MpB) functioned as hydrophilic mobile phases. The initial condition of the mobile phase was 20% MpA/80% MpB and maintained for 150 min to obtain DMP (1.2 mg), and then MpA was increased up to 50% to elute BA-I (1.7 mg). The purities of DMP and BA-I were 94.1% and 98.3% with the recovery yields of 83% and 86%, respectively. Similar results were obtained by linear-gradient CPC. The CPC peak fractions were identified by comparing their retention time to those of authentic samples of DMP and BA-I and their spectroscopic data ($^1$H NMR and $^{13}$C NMR) to those of literature values.

• 한국잡초학회지
• /
• 제17권3호
• /
• pp.303-309
• /
• 1997

가막사리 50개 수집종을 대상으로 여러 가지 형태적(形態的), 생리적(生理的)특성의 변이를 조사하고, 초형(草形), 열엽편수(裂葉片數), 수과(瘦果)길이, 개화소요일수(開花所要日數)를 기준(基準)으로 Ward의 최소분산법(最小分散法)으로 군집분석(群集分析)을 한 결과는 다음과 같다. 가막사리 수집종은 초형, 경장, 개화소요일수, 엽장폭비, 열엽편수, 엽록소함량, 엽색, 경색, 수과색, 수과길이, 수과장폭비 에서 다양한 변이를 보였으며, 수집종에 따라 약간의 변이가 있으나 대부분 상당한 휴면성과 광발아성이 있었다. 초형은 직립형(I)에서 삼각형(V)까지 다양했는데, 마름모형(III)이 가장 많았고, 경장과는 5% 수준에서 부(負)의 상관관계가 있는 것으로 나타났다. 열엽편수는 3개인 것이 가장 많았고 그 다음은 5개의 순으로 나타났는데 열엽(裂葉)되지 않는 수집종도 있었다. 수과길이 및 엽록소함량과는 5% 수준에서 엽장폭비와는 1% 수준에서 부(負)의 상관관계가 있었다. 개화소요일수는 94-141 일 범위였다. 수과길이와는 1% 수준에서, 엽장폭비와는 5% 수준에서 정(正)의 상관관계, 그리고 수과색과는 1% 수준에서 부(負)의 상관관계를 보였다. 수과길이는 6.5-14.0mm의 범위로, 평균 10.0mm, 표준편차 1.7로 9.0-12.0mm 사이가 60%를 차지했다. 수과장폭비, 경장, 개화소요일수와는 1% 수준에서, 엽장폭비와는 5% 수준에서 정(正)의 상관관계를, 그리고 엽색, 경색, 수과색, 열엽편수와는 5% 수준에서 부(負)의 상관관계를 보였다. Ward의 최소분산법을 이용한 군집분석 결과 거리 0.06에서 식물학적으로 특성이 각기 다른 6개 군으로 분류할 수 있었다.

• 분석과학
• /
• 제6권1호
• /
• pp.1-8
• /
• 1993

음이온 계면활성제인 sodium dodecyl sulfate(SDS)를 이동상으로 한 미셀 액체 크로마토그래피(micellar liquid chromatography : MLC)에서, 시료로 22종의 페놀과 벤젠 일치환체들을 선택하여 이 시료들의 용리거동을 조사하여 MLC계에서의 소수성 효과를 알아 보았다. MLC에서의 머무름과 미셀 농도간의 상관관계를 통해 시료들이 미셀-물간, 변형된 정지상-물 사이에서 분배될 때의 분배계수를 구하였고, 또 이를 이용하여 물-미셀간의 전이 자유 에너지를 구하였다. 이러한 MLC에서 얻은 소수성 파라미터들은 옥탄올-물계와 연관시켜 본 결과 좋은 상관관계를 보였다. 따라서 이 두 계에서 시료의 소수성이 머무름에 중요한 영향을 미침을 알 수 있었다. 한편, MLC를 이용하면 quantitative structure activity relationships(QSAR)에서의 소수성 정량 연구에 적용이 가능하다. 또한 알킬 치환체들에서 탄소 수의 증가에 따른 머무름 관계를 통하여 소수성의 선택성을 알아 보았고, 유기 변형제로 n-프로판올을 첨가하였을 때에도 마찬가지로 MLC계에서의 소수성 파라미터와 옥탄올-물계 사이에 상관관계가 있음을 알았다.

• Nuclear Engineering and Technology
• /
• 제31권6호
• /
• pp.572-585
• /
• 1999

The characteristics of two-phase flow instability in a vertical boiling channel connecting with an unheated riser are investigated through the linear stability analysis model. Various two-phase flow models, including thermal non-equilibrium effects, are taken into account for establishing a physical model in the time domain. A classical approach to the frequency response method is adopted for the stability analysis by employing the D-partition method. The adequacy of the linear model is verified by evaluating experimental data at high quality conditions. It reveals that the flow-pattern-dependent drift velocity model enhances the prediction accuracy while the homogeneous equilibrium model shows the most conservative predictions. The characteristics of density wave oscillations under low-power and low-quality conditions are investigated by devising a simple model which accounts for the gravitational and frictional pressure losses along the channel. The necessary conditions for the occurrences of type-I instability and flow excursion are deduced from the one-dimensional D-partition analysis. The parametric effects of some design variables on low quality oscillations are also investigated.

• 설비공학논문집
• /
• 제9권3호
• /
• pp.312-322
• /
• 1997

The natural convective flow in a two-dimensional square enclosure with horizontal partitions is investigated numerically. The enclosure was composed of the lower hot and the upper cold horizontal walls and the adiabatic vertical walls, and two identical partitions were positioned perpendicularly at the mid-height of the right and left walls, respectively. The governing equations are solved by using the finite element method with Galerkin method. Calculations are made for different partition lengths, partition conductivites, and Rayleigh numbers based on the temperature difference between two horizontal walls and the enclosure height with water(Pr=4.95). An oscillatory motion of the natural convective flow is affected significantly by the variation of the gap width and Rayleigh number. When the gap width is comparatively short, the heat transfer rate is raised with the increase of the thermal conductivity of partitions. However, for sufficiently large gap widths at higher Rayleigh numbers, the average Nusselt numbers of the conductive partitions are smaller than those of the adiabatic partitions.

• Environmental Engineering Research
• /
• 제11권1호
• /
• pp.33-44
• /
• 2006

The objective of this study was to develop and assess a method for estimating the rain scavenging ratios (RSRs) of particle-bound PAHs and PCBs using measured scavenging ratio of particulate matters (PM) and routinely available data of physico-chemical properties of PM. Paired atmospheric and rainwater sampling was conducted for a total of 4 rain events. Assuming equilibrium partitioning in rainwater-gas-PM system, an equation was derived for estimating the RSR of particle-bound chemicals as a function of RSR of PM and three equilibrium partition constants (i.e. dimensionless Henry's law constant, gas-particle partition coefficient, and water-particle partition coefficient). For all PAHs, the model significantly under-predicted the RSR while the model prediction for PCBs agreed with observation mostly within a factor of 5. The RSR model for the chemicals is of limited use as its accuracy critically depends on how close the observed partitioning of the chemicals in the gas-PM-rainwater system is to that estimated under the equilibrium assumption.

• 한국컴퓨터정보학회논문지
• /
• 제20권11호
• /
• pp.69-75
• /
• 2015

This paper suggests heuristic polynomial time algorithm for crew scheduling problem that is a kind of optimization problems. This problem has been solved by linear programming, set cover problem, set partition problem, column generation, etc. But the optimal solution has not been obtained by these methods. This paper sorts transit costs $c_{ij}$ to ascending order, and the task i and j crew paths are merged in case of the sum of operation time ${\Sigma}o$ is less than day working time T. As a result, we can be obtain the minimum number of crews $_{min}K$ and minimum transit cost $z=_{min}c_{ij}$. For the transit cost of specific number of crews $K(K>_{min}K)$, we delete the maximum $c_{ij}$ as much as the number of $K-_{min}K$, and to partition a crew path. For the 5 benchmark data, this algorithm can be gets less transit cost than state-of-the-art algorithms, and gets the minimum number of crews.

• 대한수학회지
• /
• 제45권6호
• /
• pp.1613-1622
• /
• 2008

Let G be a graph with vertex set V(G) and edge set E(G), and let g, f be two nonnegative integer-valued functions defined on V(G) such that $g(x)\;{\leq}\;f(x)$ for every vertex x of V(G). We use $d_G(x)$ to denote the degree of a vertex x of G. A (g, f)-factor of G is a spanning subgraph F of G such that $g(x)\;{\leq}\;d_F(x)\;{\leq}\;f(x)$ for every vertex x of V(F). In particular, G is called a (g, f)-graph if G itself is a (g, f)-factor. A (g, f)-factorization of G is a partition of E(G) into edge-disjoint (g, f)-factors. Let F = {$F_1$, $F_2$, ..., $F_m$} be a factorization of G and H be a subgraph of G with mr edges. If $F_i$, $1\;{\leq}\;i\;{\leq}\;m$, has exactly r edges in common with H, we say that F is r-orthogonal to H. If for any partition {$A_1$, $A_2$, ..., $A_m$} of E(H) with $|A_i|=r$ there is a (g, f)-factorization F = {$F_1$, $F_2$, ..., $F_m$} of G such that $A_i\;{\subseteq}E(F_i)$, $1\;{\leq}\;i\;{\leq}\;m$, then we say that G has (g, f)-factorizations randomly r-orthogonal to H. In this paper it is proved that every (0, mf - (m - 1)r)-graph has (0, f)-factorizations randomly r-orthogonal to any given subgraph with mr edges if $f(x)\;{\geq}\;3r\;-\;1$ for any $x\;{\in}\;V(G)$.

• International Journal of Reliability and Applications
• /
• 제12권2호
• /
• pp.103-116
• /
• 2011

A series-parallel system with independent but non-identical components is considered. The expressions have been derived for the joint reliability importance (JRI) of m (${\geq}2$) components, chosen from a series-parallel system. JRIs of components of two different series-parallel systems are studied analytically and graphically.