• Journal of the Korean Data and Information Science Society
• /
• v.28 no.6
• /
• pp.1229-1244
• /
• 2017

In recent years, as demand for data-based analytical methodologies increases in various fields, optimization methods have been developed to handle them. In particular, various constraints required for problems in statistics and machine learning can be solved by convex optimization. Alternating direction method of multipliers (ADMM) can effectively deal with linear constraints, and it can be effectively used as a parallel optimization algorithm. ADMM is an approximation algorithm that solves complex original problems by dividing and combining the partial problems that are easier to optimize than original problems. It is useful for optimizing non-smooth or composite objective functions. It is widely used in statistical and machine learning because it can systematically construct algorithms based on dual theory and proximal operator. In this paper, we will examine applications of ADMM algorithm in various fields related to statistics, and focus on two major points: (1) splitting strategy of objective function, and (2) role of the proximal operator in explaining the Lagrangian method and its dual problem. In this case, we introduce methodologies that utilize regularization. Simulation results are presented to demonstrate effectiveness of the lasso.

• The Journal of the Korea Contents Association
• /
• v.15 no.7
• /
• pp.1-11
• /
• 2015

In recent high performance computing system, GPGPU has been widely used to process general-purpose applications as well as graphics applications, since GPU can provide optimized computational resources for massive parallel processing. Unfortunately, GPGPU doesn't exploit computational resources on GPU in executing general-purpose applications fully, because the applications cannot be optimized to GPU architecture. Therefore, we provide GPU research guideline to improve the performance of computing systems using GPGPU. To accomplish this, we analyze the negative factors on GPU performance. In this paper, in order to clearly classify the cause of the negative factors on GPU performance, GPU core status are defined into 5 status: fully active status, partial active status, idle status, memory stall status and GPU core stall status. All status except fully active status cause performance degradation. We evaluate the ratio of each GPU core status depending on the characteristics of benchmarks to find specific reasons which degrade the performance of GPU. According to our simulation results, partial active status, idle status, memory stall status and GPU core stall status are induced by computational resource underutilization problem, low parallelism, high memory requests, and structural hazard, respectively.

• Journal of KIISE:Software and Applications
• /
• v.34 no.12
• /
• pp.1045-1055
• /
• 2007

Latent topic models are statistical models which automatically captures salient patterns or correlation among features underlying a data collection in a probabilistic way. They are gaining an increased popularity as an effective tool in the application of automatic semantic feature extraction from text corpus, multimedia data analysis including image data, and bioinformatics. Among the important issues for the effectiveness in the application of latent topic models to the massive data set is the efficient learning of the model. The paper proposes an accelerated learning technique for PLSA model, one of the popular latent topic models, by an incremental EM algorithm instead of conventional EM algorithm. The incremental EM algorithm can be characterized by the employment of a series of partial E-steps that are performed on the corresponding subsets of the entire data collection, unlike in the conventional EM algorithm where one batch E-step is done for the whole data set. By the replacement of a single batch E-M step with a series of partial E-steps and M-steps, the inference result for the previous data subset can be directly reflected to the next inference process, which can enhance the learning speed for the entire data set. The algorithm is advantageous also in that it is guaranteed to converge to a local maximum solution and can be easily implemented just with slight modification of the existing algorithm based on the conventional EM. We present the basic application of the incremental EM algorithm to the learning of PLSA and empirically evaluate the acceleration performance with several possible data partitioning methods for the practical application. The experimental results on a real-world news data set show that the proposed approach can accomplish a meaningful enhancement of the convergence rate in the learning of latent topic model. Additionally, we present an interesting result which supports a possible synergistic effect of the combination of incremental EM algorithm with parallel computing.

• Journal of the Korean Chemical Society
• /
• v.23 no.4
• /
• pp.217-236
• /
• 1979

The author's theory for plastic deformation was applied to superplastic alloys (Zn-Al eutectoid, Al-Cu, Pb-Sn, Sn-Bi, Mg-Al eutectics). The plastic deformation of the superplastic alloys could be described by two Maxwell models connected in parallel which represent two grain boundary flow units. The flow units are characterized by the two parameters $X_{gj}/{\alpha}_{gj}\;and\;{\beta}_{gj}$ (j=l or 2, g signifies the grain boundary) the values of which were obtained by applying our flow equation [Eq. (5)] to experiment. We confirmed that our flow equation describes the superplasticity very well. The curve of strain rate sensitivity m (=${\partial}\;In\;f/{\partial}\;In\;\dot{s})\;vs.\;-In\dot{s}$, where f and s are stress and strain rate, respectively, showed two peaks corresponding to flow unit gl and g2, the separation of the two peaks is determined by the difference between ${\beta}_{g1}\;and\;{\beta}_{g2}$. The condition of superplasticity is also determined by ${\beta}_{gj}$, which satisfies $\dot{s}_{mj}{\leqslant}1.53}{\beta}_{gj}$ [Eq.(13)], where $\dot{s}_{mj}$ is the s of the jth unit at the peak. The grain size dependence of ${\beta}_{gj}$ is described by $ln({\beta}_{gj})^{-1}$=alnx+b [Eq. (16)], where x is the grain size, and a and b are constants. The activation enthalpy for each flow unit, ${\Delta}H_{gj}^{\neq}$ was also determined from the temperature dependence of ${\beta}_{gj}$ which is proportional to the relaxation time of the j th unit. Since the superplasticity is determined by Eq. (13), and since ${\beta}_{gj}$ and ${\Delta}H_{gj}^{\neq}$ are related, we obtained the conclusion that superplasticity occurs in the system having small ${\Delta}H_{gj}^{\neq}$ values. The Aej values were equal to the activation enthalpies of grain boundary self-diffusion of the component atoms of the alloys, this accords with our proposed flow mechanism. The ${\Delta}H_{gj}^{\neq}$ value increases with grain size as expected from Eq. (16).

• Journal of Yeungnam Medical Science
• /
• v.7 no.2
• /
• pp.197-202
• /
• 1990

Mechanical obstruction of the common hepatic duct includes the following causes ; choledocholithiasis, sclerosis, cholangitis, pancreatic carcinoma, cholangiocarcinoma, postoperative stricture, primary hepatic duct carcinoma, enlarged cystic duct lymph nodes, and metastatic nodal involvement of the porta hepatis. Partial mechanical obstruction of the common hepatic duct caused by impaction of stones and inflammation surrounding the vicinity of the neck of the gallbladder had been reported on the "syndrome del conducto hepatico" in 1948 by Mirizzi. Nowadays, this disease was named by Mirizzi syndrome. Mirizzi syndrome is a rare entity of common hepatic duct obstruction that results from an inflammatory response secondary to a gallstone impacted in the cystic duct or neck of the gallbladder. It results from an almost parallel course and low insertion of the cystic duct into the common hepatic duct. In a varient of Mirizzi's syndrome, the cause of the common hepatic duct obstruction was a primary cystic duct carcinoma rather than gallstone disease. A 71-year-old man was admitted with a four-day history of right upper quadrant abdominal pain. Past medical history was unremarkable. On physical examination, the patient had a temperature of $38^{\circ}C$, icteric sclera and right upper quadrant tenderness. Pertinent laboratory findings included WBC 18,000/$cm^2$;albumin 2.6g/dl(normal 3.9-5.1) ; SGOT 183u/L(normal 0-50) ; SGPT167u/L(normal 0-65) ; bilirubin, 8.2mg/dl(normal 0-1) with the direct bilirubin, 4.4mg/dl(normal 0-0.4). Ultrasonography revealed a dilated extrahepatic biliary tree. ERCP showed that the superior margin was angular and more consistent with a calculus causing partial CHD obstruction(Mirizzi syndrome). At surgery a diseased gallbladder containing calculi was found. In addition, there was two calculi partially eroding through the proximal portion of the cystic duct and compressing the common hepatic duct. A cholecystectomy and excision of common bile duct was performed, with Roux-en-Y hepaticojejunostomy. The postoperative course was uneventful.

• The Journal of Fisheries Business Administration
• /
• v.21 no.2
• /
• pp.55-84
• /
• 1990

The objective of this study is to develop an inventory model for the inventory management of a stocking point which sells processed fisheries products. The study, first of all, sets up fisheries processing companies, food companies, apparel companies, pharmaceutical companies and electronic and electrical companies as a population. Then, a comparative study is empirically applied to obtain the inventory characteristics of final products by industry through a survey of a sample selected by a random sampling procedure. The major inventory characteristics of processed fisheries products obtained from the above analysis can be summarized as follows : 1) The major demand characteristics of processed fisheries products is to have wide seasonal fluctuations because the supply of raw materials (i.e., fisheries products) heavily depends on the productive capacity of nature. 2) It has found that fisheries processing companies are the worst in inventory management among the various industries selected in the sample. However, the self-rating of inventory management system by inventory managers of companies shows that the fisheries processing companies are relatively higher than the other companies. 3) The portion of inventory holding cost out of inventory relevant cost is very high for processed fisheries products compared with final products of the other industries. 4) Processed fisheries products are distributed to final consumers through roughly two distribution echelons and take a parallel type inventory system for their distribution structure. In order to develop an inventory model which reflects the inventory characteristics of processed fisheries products mentioned in the above, an inventory model with partial backorders is developed under the situation of stochastic lead time under the consideration of the inventory characteristics of processed fisheries products and then an iterative solution method is provided for the model. Then this study analyzes sensitivity for the standard deviation of lead time in the model by numerical examples.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.41 no.9
• /
• pp.115-124
• /
• 2004

This paper proposes a high efficient and low power FIR filter chip for partial-response maximum likelihood (PRML) disk drive read channels; it is a 6-bit, 8-tap digital FIR filter. The proposed filter employs a parallel processing architecture and consists of 4 pipeline stages. It uses the modified Booth algorithm for multiplication and compressor logic for addition. CMOS pass-transistor logic is used for low power consumption and single-rail logic is used to reduce the chip area. The proposed filter is actually implemented and the chip dissipates 120mV at 100MHz, uses a 3.3V power supply and occupies 1.88 ${\times}$ 1.38 $\textrm{mm}^2$. The implemented filter requires approximately 11.7% less power compared with the existing architectures that use the similar technology.

• Human Ecology Research
• /
• v.55 no.5
• /
• pp.491-505
• /
• 2017

This study investigated the development and core mechanism of subitizing in children and adults as well as provides related theoretical and practical discussions. This study was conducted to determine if subitizing changed with the age of participants and if there were differences in subitizing according to the spacing, color, and shape of the stimuli. The subitizing task set (including forty main trials) was prepared. Forty-five children (fifteen each in groups of 3, 4, and 5-year-olds) and fifteen adults were recruited to perform the subitizing tasks. The results demonstrated that the subitizing speed and accuracy of children improved abruptly between age 3 and age 4. Furthermore, there were significant differences in children's subitizing speed according to the spacing and color of the objects. The children's response time decreased when the objects were fully apart or were of diverse colors. In addition, there were partial significant differences in the subitizing speed of children related to the shape of the objects. The subitizing speed of children decreased in a condition (subitizing 5 objects of diverse colors in fixed spacing) when the shapes of the objects were diverse. The subitizing speed of adults only differed according to the space of the objects. The results demonstrate the development of subitizing in early childhood along with the presence of object individuation processing stages underlying subitizing. This study also provides practical information and suggestions for educational curricula that can strengthen the competency of children in systematic and diverse activities.

• Journal of Computational Design and Engineering
• /
• v.3 no.4
• /
• pp.349-362
• /
• 2016

The numerical solutions of unsteady hydromagnetic natural convection Couette flow of a viscous, incompressible and electrically conducting fluid between the two vertical parallel plates in the presence of thermal radiation, thermal diffusion and diffusion thermo are obtained here. The fundamental dimensionless governing coupled linear partial differential equations for impulsive movement and uniformly accelerated movement of the plate were solved by an efficient Finite Element Method. Computations were performed for a wide range of the governing flow parameters, viz., Thermal diffusion (Soret) and Diffusion thermo (Dufour) parameters, Magnetic field parameter, Prandtl number, Thermal radiation and Schmidt number. The effects of these flow parameters on the velocity (u), temperature (${\theta}$) and Concentration (${\phi}$) are shown graphically. Also the effects of these pertinent parameters on the skin-friction, the rate of heat and mass transfer are obtained and discussed numerically through tabular forms. These are in good agreement with earlier reported studies. Analysis indicates that the fluid velocity is an increasing function of Grashof numbers for heat and mass transfer, Soret and Dufour numbers whereas the Magnetic parameter, Thermal radiation parameter, Prandtl number and Schmidt number lead to reduction of the velocity profiles. Also, it is noticed that the rate of heat transfer coefficient and temperature profiles increase with decrease in the thermal radiation parameter and Prandtl number, whereas the reverse effect is observed with increase of Dufour number. Further, the concentration profiles increase with increase in the Soret number whereas reverse effect is seen by increasing the values of the Schmidt number.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.15 no.1
• /
• pp.245-251
• /
• 2015

In this paper, we present a design of PWM-controlled driving device for backlights in LED systems. The system can control either the brightness of the entire screen of backlights of LCD driven by LED or illumination or contrast of each partial segment of the entire screen. The PWM-controlled driving device includes the shift register that shifts the series data according to the clock signal prior to the generation of parallel data. It is also is comprised of a number of registers, a number of counters, a number of comparators, and a number of synchronizing gates (producing the PWM-controlled signals). The proposed device for backlights in LED systems can generate the PWM-controlled signal with a high degree of resolution without the increase of clock frequency. It also contains the PWM-controlled circuit that disperses and restrains the quantized noise.