• 인지과학
• /
• 제19권3호
• /
• pp.297-310
• /
• 2008

화행은 화자의 의도를 내포하기 때문에 지능형 대화 시스템을 구현할 때 화행 분석은 필수적이다. 본 논문에서는 효과적인 화행 분석을 위한 2단계 모델을 제안한다. 첫 번째 단계에서 기계 학습 기반의 신경망 모델과 확률 기반의 예측도 모델을 이용하여 개별적으로 화행 후보를 생성한다. 두 번째 단계에서는 신경망 모델이 출력한 화행 후보들을 예측도 모델이 출력한 화행 후보들로 필터링한 후, 남겨진 후보들 중에서 최대값을 가지는 화행을 선택한다. 일정관리 영역에서의 실험 결과, 제안한 2단계 모델링 방법이 기계 학습 모델이나 확률 모델만을 사용하는 기존의 방법들보다 좋은 성능을 보였다.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1994년도 춘계공동학술대회논문집; 창원대학교; 08월 09일 Apr. 1994
• /
• pp.561-568
• /
• 1994

A man model can be used as an effective tool to design ergonomically sound products and workplaces, and subsequently evaluate them properly. For a man model to be truly useful, it must be integrated with a posture prediction model which should be capable of representing the human arm reach posture in the context of equipments and workspaces. Since the human movement possesses redundant degrees of freedom, accurate representation or prediction of human movement was known to be a difficult problem. To solve this redundancy problem, a psychophysical cost function was suggested in this study which defines a cost value for each joint movement angle. The psychophysical cost function developed integrates the psychophysical discomfort of joints and the joint range availability concept which has been used for redundant arm manipulation in robotics to predict the arm reach posture. To properly predict an arm reach posture, an arm reach posture prediction model was then developed in which a posture configuration that provides the minimum total cost is chosen. The predictivity of the psychophysical cost function was compared with that of the biomechanical cost function which is based on the minimization of joint torque. Here, the human body is regarded as a two-dimensional multi-link system which consists of four links ; trunk, upper arm, lower arm and hand. Real reach postures were photographed from the subjects and were compared to the postures predicted by the model. Results showed that the postures predicted by the psychophysical cost function closely simulated human reach postures and the predictivity was more accurate than that by the biomechanical cost function.

• Bulletin of the Korean Chemical Society
• /
• 제33권2호
• /
• pp.613-619
• /
• 2012

We applied several machine learning methods for developing QSAR models for prediction of acute toxicity to fathead minnow. The multiple linear regression (MLR) and artificial neural network (ANN) method were applied to predict 96 h $LC_{50}$ (median lethal concentration) of 555 chemical compounds. Molecular descriptors based on 2D chemical structure were calculated by PreADMET program. The recursive partitioning (RP) model was used for grouping of mode of actions as reactive or narcosis, followed by MLR method of chemicals within the same mode of action. The MLR, ANN, and two RP-MLR models possessed correlation coefficients ($R^2$) as 0.553, 0.618, 0.632, and 0.605 on test set, respectively. The consensus model of ANN and two RP-MLR models was used as the best model on training set and showed good predictivity ($R^2$=0.663) on the test set.

• The Journal of Asian Finance, Economics and Business
• /
• 제9권3호
• /
• pp.181-193
• /
• 2022

This paper seeks to investigate major macroeconomic factors and bond yield interactions in Thai bond markets, with the goal of forecasting future bond yields. This study examines the best predictive yields for future bond yields at different maturities of 1-, 3-, 5-, 7-, and 10-years using time series data of economic indicators covering the period from 1998 to 2020. The empirical findings support the hypothesis that macroeconomic factors influence bond yield fluctuations. In terms of forecasting future bond yields, static predictions reveal that in most cases, the BVAR model offers the best predictivity of bond rates at various maturities. Furthermore, the BVAR model has the best performance in dynamic rolling-window, forecasting bond yields with various maturities for 2-, 4-, and 8-quarters. The findings of this study imply that the BVAR model forecasts future yields more accurately and consistently than other competitive models. Our research could help policymakers and investors predict bond yield changes, which could be important in macroeconomic policy development.

• Bulletin of the Korean Chemical Society
• /
• 제32권12호
• /
• pp.4352-4360
• /
• 2011

A four dimensional quantitative structure activity relationship analysis was applied to a series of 50 flavonoid inhibitors of $p56^{lck}$ protein tyrosine kinase by the molecular comparative electron topological method. It was found that the -log (IC50) values of the compounds were highly dependent on the topology, size and electrostatic character of the substituents at seven positions of the flavonoid scaffold in this study. Depending on the negative or positive charge of the groups correctly embedded in these substituents, three-dimensional bio-structure to increase or decrease -log (IC50) values in the training set of 39 compounds was predicted. The test set of 11 compounds was used to evaluate the predictivity of the model. To generate 4D-QSAR model, the defined function groups and pharmacophore used as topological descriptors in the calculation of activity were of sufficient statistical quality ($R^2$ = 0.72 and $Q^2$ = 0.69). Ligand docking approach by using Dock 6.0. These compounds include many flavonoid analogs, They were docked onto human families of p56lck PTKs retrieved from the Protein Data Bank, 1lkl.pdb.

• 소성∙가공
• /
• 제32권2호
• /
• pp.74-80
• /
• 2023

A deformation behavior of commercially pure titanium (CP-Ti) is highly dependent on material and processing parameters, such as deformation temperature, deformation direction, and strain rate. This study aims to predict the multivariable and nonlinear tensile behavior of CP-Ti using machine learning based on three algorithms: artificial neural network (ANN), light gradient boosting machine (LGBM), and long short-term memory (LSTM). The predictivity for tensile behaviors at the cryogenic temperature was lower than those in the room temperature due to the larger data scattering in the train dataset used in the machine learning. Although LGBM showed the lowest value of root mean squared error, it was not the best strategy owing to the overfitting and step-function morphology different from the actual data. LSTM performed the best as it effectively learned the continuous characteristics of a flow curve as well as it spent the reduced time for machine learning, even without sufficient database and hyperparameter tuning.