We compare two successful discriminative classification algorithms on three databases from the UCI and STATLOG repositories. The two approaches are the log-linear model for the class posterior probabilities and class-dependent weighted dissimilarity measures for nearest neighbor classifiers. The experiments show that the maximum entropy based log-linear classifier performs better for the equiva...

the improved estimator of the variance in the general linear model is presented under an asymmetric linex loss function.

In mixtures-of-experts (ME) models, "experts" of generalized linear models are combined, according to a set of local weights called the "gating function". The invariant transformations of the ME probability density functions include the permutations of the expert labels and the translations of the parameters in the gating functions. Under certain conditions, we show that the ME systems are iden...

Varying-coeecient models are an extension of generalized linear models. In this case the coeecients are allowed to vary smoothly with the value of other variables. We present a "local Fisher scoring backktting" algorithm based on a weighted local likelihood approach where the components are iteratively tted. An example will illustrate the usefulness of this approach.

Generalized F tests were introduced by Michalski and Zmyślony (1996) for variance components and later (1999) for linear functions of parameters in mixed linear models. We now use generalized polar coordinates to obtain, for the second case, tests that are more powerful for selected families of alternatives.

By using themethods of linear algebra andmatrix inequality theory, we obtain the characterization of admissible estimators in the general multivariate linear model with respect to inequality restricted parameter set. In the classes of homogeneous and general linear estimators, the necessary and suffcient conditions that the estimators of regression coeffcient function are admissible are establi...

the fuzzy linear regression model with fuzzy input-output data andcrisp coefficients is studied in this paper. a linear programmingmodel based on goal programming is proposed to calculate theregression coefficients. in contrast with most of the previous works, theproposed model takes into account the centers of fuzzy data as animportant feature as well as their spreads in the procedure ofconstr...

In this paper, a new method for estimating the linear regression coefficients approximation is presented based on Z-numbers. In this model, observations are real numbers, regression coefficients and dependent variables (y) have values ​​for Z-numbers. To estimate the coefficients of this model, we first convert the linear regression model based on Z-numbers into two fuzzy linear regression mode...

the stability analysis of islanded inverter-based microgrids (ibmgs) is increasingly an important and challenging topic due to the nonlinearity of ibmgs. in this paper, a new linear model for such microgrids as well as an iterative method to correct the linear model is proposed. using the linear model makes it easy to analyze the eigenvalues and stability of ibmgs due to the fact that it derive...

Abstract: We discuss a class of graphical models for discrete data defined by what we call a multivariate regression chain graph Markov property. We propose a parameterization based on a sequence of generalized linear models with a multivariate logistic link function. We show the relationship with a chain graph model recently defined in the literature, and we prove that the proposed parametriza...