Search Results

You are looking at 1 - 10 of 33 items :

  • "model identifiability" x
Clear All

) are a class of graphical models introduced to analyze the results of gene perturbation screens. NEMs explore noisy subset relations between the high- dimensional outputs of phenotyping studies, e.g., the effects showing in gene expression profiles or as morphological features of the perturbed cell. In this paper we expand the statistical basis of NEMs in four directions. First, we derive a new formula for the likelihood function of a NEM, which generalizes previous results for binary data. Second, we prove model identifiability under mild assumptions. Third, we show

{)}+(ib+jc+kd)% \Bigl{(}1-\frac{1}{|\alpha|^{2}}\Bigr{)}\Bigr{)}, Lemmas 3.2 and 3.3 are also valid for quaternion random variables, and we use Theorem 4.1 instead of Theorem 1.1 . ∎ References [1] J. Eriksson and V. Koivunen, Complex random vectors and ICA models: Identifiability, uniqueness, and separability, IEEE Trans. Inform. Theory 52 (2006), no. 3, 1017–1029. 10.1109/TIT.2005.864440 Eriksson J. Koivunen V. Complex random vectors and ICA models: Identifiability, uniqueness, and separability IEEE Trans. Inform. Theory 52 2006 3 1017 1029 [2] G. M. Fel

performance of the two considered teams on the group stage, additional parameters are introduced: 4 Journal of Quantitative Analysis in Sports, Vol. 7 [2011], Iss. 1, Art. 6 DOI: 10.2202/1559-0410.1275 λ1(i) specifies the main effect of team T1 being a group winner (i = 1) or second- place finisher (i = 2). In an analogue way λ2( j) is the main effect of the success of team T2 on the group stage ( j = 1,2). Finally, interactions of the group stage indicators G1 (team T1) and G2 (team T2) are modeled via λ1,2(i, j). To make the model identifiable, symmetric restrictions

, characterization of probability distri- bution, beta conditional distribution, conditional moments, linearity of regression, gamma distribution, beta type distributions. 72 A.K. Gupta, J. Wesoiowski (1995a, b, 1996a, b), Gupta and Wesoiowski (1997, 1999), Papageorgiou and Wesoiowski (1997). In this paper we are interested in beta mixtures with a mixing variable being the right end of the support. A uniform mixture is a special kind of such a model. Identifiability and identification problems for uniform mixtures were considered, using other methods, in Gupta and

- ment prétendre tenir un discours sur l’événement. Une comparaison des photo- graphies prises par un témoin à la suite des attentats avec celles des reporters, tend à les rassembler toutes sous un modèle unique dans la mesure où elles s’attachent à l’acmé de l’action, à l’instant décisif censé condenser la totalité de l’événement dont il est extrait. En dépit de cette ressemblance superficielle, les images font apparaître deux modèles identifiables, deux textualités diffé- rentes et reconnaissables qui laissent à l’observateur une “impression de déjà vu” (Eco

error term with mean zero and variance σ ε 2 $$\sigma _\varepsilon ^2$$ . In order to make the model identifiable, one needs to impose the following constraints on the parameters, i.e. [3] ∑ t = t 1 t n k t = 0 , a n d ∑ x = x 1 x p β x = 1. $$\mathop \sum \limits_{t = {t_1}}^{{t_n}} {k_t} = 0,\,{\rm{and}}\,\mathop \sum \limits_{x = {x_1}}^{{x_p}} {\beta _x} = 1.$$ 3 Estimation Parameter Approaches 3.1 Singular Value Decomposition (SVD) In its original version, the Lee-Carter model cannot fit into ordinary regression methods because there are no given regressors on

rate of fatigue dosage, in terms of mph lost, begins to noticeably increase after 15 pitches and rises substantially after 21 pitches. We choose to model the pitcher-level intercepts α j and response magnitudes m j hierarchically, fixing the global mean of m j at 1 in order to make the model identifiable. As such, we specify the diffuse priors (7) α j ∼ Normal ( μ α , τ α 2 )    μ α ∼ Normal ( 93 , 3 2 ) τ α ∼ HC ( 0 , 1 ) $$\begin{array}{lllll}&\alpha_{j}\sim\text{Normal}(\mu_{\alpha},\tau^{2}_{\alpha})\qquad\mu_{\alpha}\sim\text{Normal}(93,3^{2})\\ &\tau

-algebra based techniques have been employed for differential algebraic as well as ordinary differential equations systems. Fliess and coworkers (Diop, 2001; Fliess, 1990) develop much of the work in the control literature on the employ of differential-algebra. Important contributions of these techniques have been used for model identifiably (Fliess, et al., 1999). In order to give a background previous to the estimation methodology proposed under the differential algebraic frame, the following definitions are considered: Definition 1 Let L and K be differential

Vk is a factor loading matrix, and therefore the resulting model is not strictly identifiable since Vk can be rotated without affecting the results, provided that the latent vectors uk are counter-rotated. On the other hand, we assume that the matrix Bk is binary and row stochastic and this helps making the model identifiable. In fact, let T be an arbitrary Qk × Qk orthogonal matrix such that TT′ = T′T = IQk . Then, consider the orthogonal transformation B∗k = BkT. The transformed class-conditional covariance matrix is equal BkB ′ k + Dk. In fact, the covariance

, where no validation data were available and covariates were subject to high levels of missingness according to a missing not at random pattern. Such a framework has the potential to be expanded to also include error-prone or misclassified covariate data. Important challenges for settings with more complex error structures may remain for this approach; in particular, model identifiability is a known challenge for latent class models ( Gustafson 2005 ) and model identifiability for latent class models with misclassified outcomes is an active area of research for both