• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.9
• /
• pp.2505-2528
• /
• 2023

The field of facial expression recognition (FER) has been actively researched to improve human-computer interaction. In recent years, deep learning techniques have gained popularity for addressing FER, with numerous studies proposing end-to-end frameworks that stack or widen significant convolutional neural network layers. While this has led to improved performance, it has also resulted in larger model sizes and longer inference times. To overcome this challenge, our work introduces a novel lightweight model architecture. The architecture incorporates three key factors: Depth-wise Separable Convolution, Residual Block, and Attention Modules. By doing so, we aim to strike a balance between model size, inference speed, and accuracy in FER tasks. Through extensive experimentation on popular benchmark FER datasets, our proposed method has demonstrated promising results. Notably, it stands out due to its substantial reduction in parameter count and faster inference time, while maintaining accuracy levels comparable to other lightweight models discussed in the existing literature.

• The Korean Journal of Applied Statistics
• /
• v.30 no.2
• /
• pp.203-213
• /
• 2017

We study estimation and inference of joint conditional distributions of bivariate longitudinal outcomes using regression models and copulas. We consider a class of time-varying transformation models and combine the two marginal models using Gaussian copulas to estimate the joint models. Our models and estimation method can be applied in many situations where the conditional mean-based models are inadequate. Gaussian copulas combined with time-varying transformation models may allow convenient and easy-to-interpret modeling for the joint conditional distributions for bivariate longitudinal data. We apply our method to an epidemiological study of repeatedly measured bivariate cholesterol data.

• Asian Pacific Journal of Cancer Prevention
• /
• v.15 no.9
• /
• pp.4049-4054
• /
• 2014

Background: Race and ethnicity are significant factors in predicting survival time of breast cancer patients. In this study, we applied advanced statistical methods to predict the survival of White non-Hispanic female breast cancer patients, who were diagnosed between the years 1973 and 2009 in the United States (U.S.). Materials and Methods: Demographic data from the Surveillance Epidemiology and End Results (SEER) database were used for the purpose of this study. Nine states were randomly selected from 12 U.S. cancer registries. A stratified random sampling method was used to select 2,000 female breast cancer patients from these nine states. We compared four types of advanced statistical probability models to identify the best-fit model for the White non-Hispanic female breast cancer survival data. Three model building criterion were used to measure and compare goodness of fit of the models. These include Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC), and Deviance Information Criteria (DIC). In addition, we used a novel Bayesian method and the Markov Chain Monte Carlo technique to determine the posterior density function of the parameters. After evaluating the model parameters, we selected the model having the lowest DIC value. Using this Bayesian method, we derived the predictive survival density for future survival time and its related inferences. Results: The analytical sample of White non-Hispanic women included 2,000 breast cancer cases from the SEER database (1973-2009). The majority of cases were married (55.2%), the mean age of diagnosis was 63.61 years (SD = 14.24) and the mean survival time was 84 months (SD = 35.01). After comparing the four statistical models, results suggested that the exponentiated Weibull model (DIC= 19818.220) was a better fit for White non-Hispanic females' breast cancer survival data. This model predicted the survival times (in months) for White non-Hispanic women after implementation of precise estimates of the model parameters. Conclusions: By using modern model building criteria, we determined that the data best fit the exponentiated Weibull model. We incorporated precise estimates of the parameter into the predictive model and evaluated the survival inference for the White non-Hispanic female population. This method of analysis will assist researchers in making scientific and clinical conclusions when assessing survival time of breast cancer patients.

• Proceedings of the KIEE Conference
• /
• 2004.11c
• /
• pp.355-357
• /
• 2004

In this paper, to calculate accurate remaining volume, it presents how to figure out nickel-cadmium battery algorithm. A nickel-cadmium battery has widely been used in industrial field and to military. Recent high demands on the battery caused 'How to calculate accurate remaining volume is very important task to be solved. In this paper, it says it is useful using the terminal voltage change of the resistance that can be connected with the battery and the differentiation of the terminal voltage to calculate remaining volume of nickel-cadmium battery. And these can be used for volume inference data so that it is fuzzy based system which can be helpful to inference the remaining volume by the resistance of terminal voltage change. Because of electrochemical complexity, the volume calculating system is inferencing undirectly by experimentally built DB where as current the existing volume models are suffering to be adapted.

• Communications for Statistical Applications and Methods
• /
• v.10 no.2
• /
• pp.341-352
• /
• 2003

Collecting data through experiment, the observers are an import source of measurement error and the inference on the measure of agreement, say kappa, is necessary. The models commonly used are complicated general mixture model, which have many nuisance parameters. Orthogonalization of parameters reduce the effect of nuisance parameter. Orthogonalization of estimating function gives the same effect as the parameter orthogonalization. In this study, the method for orthogonalization of estimating equation is studied and applied to the Beta-binomial model to examine the properties of the estimate of kappa. As a result, the likelihood function is insensitive to the change of the nuisance parameter and bias is smaller than the result of m.1.e. when kappa has extreme values

• The Korean Journal of Applied Statistics
• /
• v.13 no.2
• /
• pp.505-514
• /
• 2000

A Markov Chain Monte Carlo method with data augmentation is developed to compute the features of the posterior distribution. For each observed failure epoch, we introduced mixture failure model of Rayleigh and Erlang(2) pattern. This data augmentation approach facilitates specification of the transitional measure in the Markov Chain. Gibbs steps are proposed to perform the Bayesian inference of such models. For model determination, we explored sum of relative error criterion that selects the best model. A numerical example with simulated data set is given.

• ETRI Journal
• /
• v.26 no.4
• /
• pp.297-306
• /
• 2004

A plasma is a collection of charged particles and on average is electrically neutral. In fabricating integrated circuits, plasma etching is a key means to transfer a photoresist pattern into an underlayer material. To construct a predictive model of plasma-etching processes, a polynomial neural network (PNN) is applied. This process was characterized by a full factorial experiment, and two attributes modeled are its etch rate and DC bias. According to the number of input variables and type of polynomials to each node, the prediction performance of the PNN was optimized. The various performances of the PNN in diverse environments were compared to three types of statistical regression models and the adaptive network fuzzy inference system (ANFIS). As the demonstrated high-prediction ability in the simulation results shows, the PNN is efficient and much more accurate from the point of view of approximation and prediction abilities.

• 제어로봇시스템학회:학술대회논문집
• /
• 1994.10a
• /
• pp.379-384
• /
• 1994

In this paper, a fish-drying control method is proposed, which utilizes prediction of proper change in- weight of material fish based on skilled worker's performance. The function of the proposed system is largely broken down into two procedures: The procedure before drying and the one during drying. The procedure before drying is the determination of necessary drying conditions and the required drying time. Required drying time and proper changes in weight for a specific product are obtained by using fuzzy inference and regression models. The procedure during drying is the prediction of the state of material at the end of drying, or the state of product and regulation of drying conditions to attain the prescribed goal before drying. The prediction of product is obtained by using a set of linear-differential equations obtained by the authors' previous work. Drying conditions are regulated by using fuzzy inference. A good agreement between the results of simulation and experiments is obtained, which implies the usefulness of the present control method.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1892-1896
• /
• 2005

In this paper, an effective method for selecting significant input variables in building ANFIS (Adaptive Neuro-Fuzzy Inference System) for nonlinear system modeling is proposed. Dominant inputs in a nonlinear system identification process are extracted by evaluating the performance index and they are applied to ANFIS. The availability of our proposed model is verified with the Box and Jenkins gas furnace data. The comparisons with other methods are also given in this paper to show our proposed method is superior to other models.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.6 no.2
• /
• pp.138-143
• /
• 2006

In this paper, we have presented a Sequential Agglomerative Hierarchical Nested (SAHN) algorithm-based data clustering method in fuzzy inference system to achieve optimal performance of fuzzy model. SAHN-based algorithm is used to give possible range of number of clusters with cluster centers for the system identification. The axes of membership functions of this fuzzy model are optimized by using cluster centers obtained from clustering method and the consequence parameters of the fuzzy model are identified by standard least square method. Finally, in this paper, we have observed our model's output performance using the Box and Jenkins's gas furnace data and Sugeno's non-linear process data.