• Journal of Software Assessment and Valuation
• /
• v.17 no.1
• /
• pp.101-113
• /
• 2021

In this paper, we introduce a speaker overlap system and look at the process of converting the existed system on the specific framework of artificial intelligence. Speaker overlap is when two or more speakers speak at the same time during a conversation, and can lead to performance degradation in the fields of speech recognition or speaker recognition, and a lot of research is being conducted because it can prevent performance degradation. Recently, as application of artificial intelligence is increasing, there is a demand for switching between artificial intelligence frameworks. However, when switching frameworks, performance degradation is observed due to the unique characteristics of each framework, making it difficult to switch frameworks. In this paper, the process of converting the speaker overlap detection system based on the Keras framework to the pytorch-based system is explained and considers components. As a result of the framework switching, the pytorch-based system showed better performance than the existing Keras-based speaker overlap detection system, so it can be said that it is valuable as a fundamental study on systematic framework conversion.

• Korean Journal of Mineralogy and Petrology
• /
• v.35 no.3
• /
• pp.273-281
• /
• 2022

Mantle heterogeneity is closely related to the distribution and circulation of volatile components in the Earth's interior, and the behavior of volatiles in the mantle strongly influences the rheological properties of silicate rocks. In mantle xenoliths, these physicochemical properties of the upper mantle can be recorded in the form of microstructures and fluid inclusions. In this paper, I summarized and reviewed the results of previous studies related to the characteristics of microstructures and fluid inclusions from peridotite xenoliths beneath the Rio Grande Rift (RGR) in order to understand the evolution and heterogeneity of upper mantle. In the RGR, the mantle peridotites are mainly reported in the rift axis (EB: Elephant Butte, KB: Kilbourne Hole) and rift flank (AD: Adam's Diggings) regions. In the case of the former (EB and KB peridotites), the type-A lattice preferred orientation (LPO), formed under low-stress and low-water content, was reported. In the case of the latter (AD peridotites), the type-C LPO, formed under low-stress and high-water content, was reported. In particular, in the case of AD peridotites, at least two fluid infiltration events, such as early (type-1: CO₂-N₂) and late (type-2: CO₂-H₂O), have been recorded in orthopyroxene. The upper mantle heterogeneity recorded by these microstructures and fluid inclusions is considered to be due to the interaction between the North American plate and the Farallon plate.

• Journal of Korea Water Resources Association
• /
• v.57 no.2
• /
• pp.73-85
• /
• 2024

Predicting water quality of rivers and reservoirs is necessary for the management of water resources. Artificial Neural Networks (ANNs) have been used in many studies to predict water quality with high accuracy. Previous studies have used Gradient Descent (GD)-based optimizers as an optimizer, an operator of ANN that searches parameters. However, GD-based optimizers have the disadvantages of the possibility of local optimal convergence and absence of a solution storage and comparison structure. This study developed improved optimizers to overcome the disadvantages of GD-based optimizers. Proposed optimizers are optimizers that combine adaptive moments (Adam) and Nesterov-accelerated adaptive moments (Nadam), which have low learning errors among GD-based optimizers, with Harmony Search (HS) or Novel Self-adaptive Harmony Search (NSHS). To evaluate the performance of Long Short-Term Memory (LSTM) using improved optimizers, the water quality data from the Dasan water quality monitoring station were used for training and prediction. Comparing the learning results, Mean Squared Error (MSE) of LSTM using Nadam combined with NSHS (NadamNSHS) was the lowest at 0.002921. In addition, the prediction rankings according to MSE and R² for the four water quality indices for each optimizer were compared. Comparing the average of ranking for each optimizer, it was confirmed that LSTM using NadamNSHS was the highest at 2.25.

• Animal Systematics, Evolution and Diversity
• /
• v.29 no.1
• /
• pp.23-30
• /
• 2013

This study reports the discovery of two oxytrichid ciliates, Cyrtohymena primicirrata (Berger and Foissner, 1987) and Oxytricha granulifera Foissner and Adam, 1983, in Jeju Island, Korea. The morphology of the two species was studied using live observation and protargol impregnation. These species are described as follows: Cyrtohymena primicirrata has a body size in live specimens $90-140{\times}40-60{\mu}m$, length : width ratio 2.3 : 1 on average; elongated and slender obovate in outline of body. Cortical granules are shiny yellow on the ventral and dorsal side. Adoral zone of membranelles (AZM) is covering about 48% of the cell with about 38 adoral membranelles. Arrangement of undulating membranes is ordinary Cyrtohymena pattern. Dorsal kineties is six rows with $5{\mu}m$ long bristles. Oxytricha granulifera has a body size in live specimens $90-115{\times}25-38{\mu}m$, length : width ratio 3.31 on average; elongated ellipsoidal in outline of body. Cortical granules are colorless on the ventral and dorsal side. AZM is covering 28% of the cell length in vivo with about 24 adoral membranelles. Arrangement of undulating membranes is Oxytricha pattern. Dorsal kineties is five rows with about $3{\mu}m$ long dorsal bristles.

• Safety and Health at Work
• /
• v.6 no.2
• /
• pp.104-113
• /
• 2015

Background: Given the long hours on the road involving multiple and interacting work stressors (i.e., delivery pressures, irregular shifts, ergonomic hazards), commercial drivers face a plethora of health and safety risks. Researchers goal was to determine whether and to what extent long-haul trucker work schedules influence sleep duration and quality. Methods: Survey and biometric data collected from male long-haul truck drivers at a major truckstop in central North Carolina over a six month period. Results: Daily hours worked (mean = 11 hours, 55 minutes) and frequency of working over government-mandated daily HOS regulations (23.8% "frequently or always") were statistically significant predictors of sleep duration. Miles driven per week (mean = 2,812.61), irregular daily hours worked (63.8%), and frequency of working over the daily hour limit (23.8% "frequently or always") were statistically significant predictors of sleep quality. Conclusion: Implications of findings suggest a comprehensive review of the regulations and operational conditions for commercial motor vehicle drivers be undertaken.

• 한국HCI학회:학술대회논문집
• /
• 2006.02a
• /
• pp.75-81
• /
• 2006

슬관절 전치환술은 관절염이나 사고로 인해 일상적인 활동의 제약을 받는 환자의 슬관절을 인공 관절로 대체함으로써 본래의 기능을 복원하고자 하는 수술이다. 이 수술은 인공 관절의 위치 및 정렬에 매우 민감하게 영향을 받기 때문에 수술이 잘못되는 경우 정렬 이상으로 인한 해리, 삽입물의 파손, 인공 슬관절 주위 골절, 슬개골 탈구, 굴곡 각도의 제한 등의 증상이 발생할 수 있다. 현재의 인공 관절은 임상에 적용되는 다양한 인공 관절 중에서 적당한 형상의 관절을 선택하여 시술되고 있지만 환자의 골 형상에 정확히 일치하는 인공 관절 선택의 어려움 때문에 종종 시술 후 부작용이 발생한다든지 심지어는 재수술을 해야 될 경우도 발생하게 된다. 본 논문은 Mechanical CAD 소프트웨어인 CATIA에서 제공하는 절단, Assembly, Analysis, Kinematic Simulation 기능 등을 이용하여 가상 수술을 수행하는 과정을 보여준다. 슬관절 전치환술 과정을 그대로 재현하여 절단량과 절단각을 결정하고 환자의 골격 형상에 적합한 최적의 인공 관절을 실제 수술 전에 미리 선정할 수 있다. CAD 시스템을 이용함으로써 외과의들이 실제 수술 시에 시행착오법을 통해 인공 관절을 선택하는 과정을 줄이고 수술의 정확도를 높일 수 있다. 향후 ADAMS나 ANSYS와 연계하여 수술 후 동작이나 하중을 분석할 수 있으며, 수술 과정에 대한 교육용으로 활용될 수 있다.

• The Journal of Asian Finance, Economics and Business
• /
• v.7 no.12
• /
• pp.433-444
• /
• 2020

The purpose of this study is to analyze integrated reporting disclosure and its implications on investor reactions. The population in this study is all manufacturing companies listed on the Indonesia stock exchange from 2017 to 2019, totaling 171 companies, and the sampling technique used is purposive sampling method. The method used in this research is a quantitative description using the financial statements of manufacturing companies listed on the Indonesia stock exchange. The data analysis method used is multiple regression analysis with intervening variables using AMOS 24 software. The results of this study show a positive and significant effect of profitability (X1) and company size (X2) on integrated reporting (IR); a positive and insignificant effect of stakeholder pressure (X3) on integrated reporting (IR); a positive and significant effect of profitability (X1) and stakeholder pressure (X3) on investor reactions (Y); a positive and insignificant effect of firm size (X2) and integrated reporting (IR) on investor reactions (Y). Suggestions are that in further studies, we can increase the sample size by including other industries, and in addition to using annual reporting, we can also use other sources such as websites, press releases, and prospectuses to improve the robustness of this study by relying on other data sources.

• Journal of Korea Multimedia Society
• /
• v.23 no.9
• /
• pp.1129-1138
• /
• 2020

The performance of convolutional deep learning networks is generally determined according to parameters of target dataset, structure of network, convolution kernel, activation function, and optimization algorithm. In this paper, a genetic algorithm is used to select the appropriate deep learning model and parameters for Alzheimer's classification and to compare the learning results with preliminary experiment. We compare and analyze the Alzheimer's disease classification performance of VGG-16, GoogLeNet, and ResNet to select an effective network for detecting AD and MCI. The simulation results show that the network structure is ResNet, the activation function is ReLU, the optimization algorithm is Adam, and the convolution kernel has a 3-dilated convolution filter for the accuracy of dementia medical images.

• Smart Structures and Systems
• /
• v.19 no.2
• /
• pp.225-236
• /
• 2017

Railroad bridges in North America are an integral but aging part of the railroad network and are typically only monitored using visual inspections. When quantitative information is required for assessment, railroads often monitor bridges using accelerometers. However without a sensor to directly measure displacements, it is difficult to interpret these results as they relate to bridge performance. Digital Image Correlation (DIC) is a non-contact sensor technology capable of directly measuring the displacement of any visible bridge component. In this research, a railroad bridge was monitored under load using DIC and accelerometers. DIC measurements are directly compared to serviceability limits and it is observed that the bridge is compliant. The accelerometer data is also used to calculate displacements which are compared to the DIC measurements to assess the accuracy of the accelerometer measurements. These measurements compared well for zero-mean lateral data, providing measurement redundancy and validation. The lateral displacements from both the accelerometers and DIC at the supports were then used to determine the source of lateral displacements within the support system.

• BMB Reports
• /
• v.43 no.5
• /
• pp.375-381
• /
• 2010

In this study, the cDNA library of Chang-liver cells was immunoscreened using common ADAMs antibody to obtain ADAM related genes. We found one positive clone that was confirmed as a new gene by Blast, which is an uncharacterized helical and coil protein and processes protease activity, and named protease-related protein 1 (ARP1). The submitted GenBank accession number is AY078070. Molecular characterizations of ARP1 were analyzed with appropriate bioinformatics software. To analyse its expression and function, ARP1 was subcloned into glutathione S-transferase fusion plasmid pGEX-2T and expressed by E. coli system. The in vitro expression product of ARP1 was recognized by common ADAMs antibody with western blot. Interestingly, ARP1 cleaves gelatine at pH9.5, which suggests it is an alkaline protease. Semi-quantitative RT-PCR result indicates that ARP1 mRNA is strongly transcribed in the liver and the treated Chang-liver cells.