S. E. Fienberg, The Analysis of Cross--Classified Categorical Data, vol. Second Edition. MIT Press, 1981.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of the image lattice. This is a much stronger assumption than made for a normal MRF which defines a site as being conditionally independent upon its non neighbouring sites given all of the neighbouring sites. This strong MRF model is equivalent to the Analysis of variance (ANOVA) construction [2, 7], which allows us to use the theorems from the ANOVA construction to estimate the LCPDF for the strong MRF model. 3. Multiscale texture synthesis To synthesis a texture we used our multiscale relaxation (MR) algorithm as formalised in [8] The basis of the algorithm is to perform stochastic ....

S. E. Fienberg. The Analysis of Cross--Classified Categorical Data, volume Second Edition. MIT Press, 1981.

....of independence even when there is no basis for the assumption. Consequently, we have assumed an extra degree of conditional independence so as to simplify the MRF model to a strong MRF model. By demonstrating the equivalence between the strong MRF model and the ANOVA log linear construction [3] [6], we are able to use the estimation processes developed for the ANOVA log linear construction to calculate the probability distribution associated with the strong MRF model. In return, Moussouris s [12] strong MRF formula gives the general formula for the ANOVA log linear construction. II. ....

....mathematical constructions. The first proof presented in Appendix A, is based on the similar proof by Grimmett [10] and Moussouris [12] for the equivalence of a standard MRF and a Gibbs distribution. The second proof presented in Appendix B, is based on the ANOVA log linear construction [3] [6] for testing independence in a distribution. As both mathematical constructions are used to prove Proposition 1, the constructions are equivalent in terms of the strong MRF. IV. ESTIMATION OF THE STRONG LCPDF A. Direct estimate technique Bishop et al. 3] did not derive the direct estimate Eq. ....

[Article contains additional citation context not shown here]

S. E. Fienberg. The Analysis of Cross--Classified Categorical Data, volume Second Edition. MIT Press, 1981.

....thereby producing a model suitable for open ended classification of texture in an image. This second model is based on Moussouris strong MRF model [148] In this thesis we show that Moussouris strong MRF model is equivalent to the Analysis of variance 1.6. OUTLINE OF THESIS 19 (ANOVA) model [67], which allows us to use the theorems of the ANOVA model for estimating the strong MRF model. For the open ended classification of texture, we use our strong MRF model to find the lowest order statistics that may be used to uniquely represent the texture. The model is then used to collect a ....

....For this we introduce the strong MRF model. With it we can adjust the order of the statistics used to model the texture. This second model is based on Moussouris strong MRF model [148] In this thesis we show that Moussouris strong MRF model is equivalent to the Analysis ofvariance (ANOVA) model [67], which allows us to use the theorems of the ANOVA model [22] for the strong MRF model. For the open ended classification of texture, we use our strong MRF model to find the lowest order statistics that may be used to uniquely represent the texture. The model is then used to collect a sample of ....

[Article contains additional citation context not shown here]

S. E. Fienberg, The Analysis of Cross--Classified Categorical Data, vol. Second Edition. MIT Press, 1981.

....subset of S. This is a much stronger assumption than made for a normal MRF which defines a site as being conditionally independent upon its non neighbouring sites given all of the neighbouring sites. We show that this strong MRF model is equivalent to the Analysis ofvariance (ANOVA) construction [18]. This equivalence allows us to use the theorems from the ANOVA construction to estimate the LCPDF of the strong MRF model. MRF condition . 14) Strong MRF condition s, r S) P (x s , A The strong MRF condition states that the LCPDF # s (x s S) ....

....from the local clique set C . We provide two proofs for Proposition 1. The first method relies on the Mobius inversion formula (4) and follows Grimmett s [25] and Moussouris s [36] construction for the potential V , Appendix A. The second proof is based on the ANOVA construction [18] for testing variable independence in a distribution, Appendix B. The proofs show that the strong MRF model is equivalent to the ANOVA construction. Even though equation (22) represents the general clique decomposition for P (x s , x r , r s ) it is subject to condition (43) Bishop et al. ....

[Article contains additional citation context not shown here]

Stephen E. Fienberg, The Analysis of Cross--Classified Categorical Data, vol. Second Edition, MIT Press, 1981.

....of the image lattice. This is a much stronger assumption than made for a normal MRF which defines a site as being conditionally independent upon its non neighbouring sites given all of the neighbouring sites. This strong MRF model is equivalent to the Analysis of variance (ANOVA) construction [3], which allows us to use the theorems from the ANOVA construction to estimate the LCPDF for the strong MRF model. The ability to use a strong MRF model allowed us to not only to vary the neighbourhood size, but also the statistical order of the model. In the classification analysis, we were able ....

S. E. Fienberg. The Analysis of Cross--Classified Categorical Data, volume Second Edition. MIT Press, 1981.

....independence so as to simplify the MRF model to a strong MRF model. This simplification allows the model to be defined directly from the random field s marginal distributions. It also allows an equivalence to be shown between the strong MRF model and the Analysis of variance (ANOVA) construction [4, 9]. By demonstrating the equivalence between the strong MRF model and the Analysis of variance (ANOVA) construction [4, 9] two scientific avenues have their field of knowledge expanded. For the strong MRF model, the estimation processes developed for the ANOVA construction can be used to calculate ....

....directly from the random field s marginal distributions. It also allows an equivalence to be shown between the strong MRF model and the Analysis of variance (ANOVA) construction [4, 9] By demonstrating the equivalence between the strong MRF model and the Analysis of variance (ANOVA) construction [4, 9], two scientific avenues have their field of knowledge expanded. For the strong MRF model, the estimation processes developed for the ANOVA construction can be used to calculate the probability distribution associated with the strong MRF model. However, an even more beneficial outcome is that ....

[Article contains additional citation context not shown here]

S. E. Fienberg. The Analysis of Cross--Classified Categorical Data, volume Second Edition. MIT Press, 1981.

....The responses of the subjects were pooled into a threedimensional table with combination of pitch level and pitch range as one dimension, intonation pattern as a second dimension, and response of the subjects as a third one. These data were subjected to a three way log linear analysis [5]. In the simplest log linear model into which the data can be fitted, there were significant interactions between pitch level pitch range combination and response , and between intonation pattern and response , but not between pitch level pitch range combination and intonation ....

Fienberg, S. E. 1980. The analysis of cross-classified categorical data, second edition. The MIT Press, Cambridge, Massachusetts.

....fit for the data and removing a (k 1) way interaction from a corresponding log linear model does not affect the fitness of the model. In general, Pearson s goodness of fit chisquare statistic is recommended for its relatively stable approximate to the c 2 distribution with small sample sizes (Fienberg, 1980). Markov models of the third or lower order are examined to determine the appropriate order of the Markov chains. Previous research has shown that search patterns with online information systems can be adequately modelled by a second order Markov chain (e.g. Qiu, 1993) The results of this ....

Fienberg, S. E. (1980) The analysis of cross-classified categorical data. Cambridge, Mass.: The MIT Press.

....Discovered clusters are unions of adjacent high density units. 2.4 Correlation Statistically oriented in nature, correlation has seen increasing use as a data mining technique. Although the analysis of multi dimensional categorical data is possible and described extensively in the literature [56, 16, 22], the most commonly employed method is that of two dimensional contingency table analysis of categorical data using the chi square statistic as a measure of significance. Recent examples from the literature include the work of Sanjeev and Zytkow [58] Knobbe and Adrian [35] Zemobowicz and Zytkow ....

S.E. Fienberg. The analysis of cross-classified categorical data. MIT Press, 1978.

.... articles from the Penn Treebank (Marcus et al. 1993) deriving collocational information of this type (this work is currently unpublished) Statistically motivated approaches such as the likelihood ratio test (Dunning 1993) mutual information (Church et al. 1991) and the c 2 test (Hoel 1971; Fienberg 1977), could also be used to identify domain specific terms and collocations occurring within the relevant documents. This will enable concentration of the technique on terms for which there is a higher probability of relevance. One class of feature slots used in the MUC competitions that is not ....

Fienberg, S.E. 1977. The Analysis of Cross-classified Categorical Data. Cambridge, Massachusetts.: MIT Press.

....produced employing the cloglog link and the functions gam( and predict:gam( of the statistical package S (see[2, 8] The data are provided in an Appendix. 4 Ordinal Data. A number of different models have been proposed for the analysis of ordinal data. These include: continuation ratio (see [12]) stereotype (see [1] and the grouped continuous (see [20] The following presents an approach to building a stochastic model for ordinal data. Let Y be 0; 1; 2 for a particular game, depending on whether the result is a loss; tie or win. Suppose that there exists a latent or state variable, ....

S.E. Fienberg. The Analysis of Cross-Classified Categorical Data. MIT Press, Cambridge, Mass., 1980.

....subset of S. This is a much stronger assumption than made for a normal MRF which defines a site as being conditionally independent upon its non neighbouring sites given all of the neighbouring sites. We show that this strong MRF model is equivalent to the Analysis ofvariance (ANOVA) construction [18]. This equivalence allows us to use the theorems from the ANOVA construction to estimate the LCPDF of the strong MRF model. MRF condition Pi s (x s jx r ; r 6= s) P (x s jx r ; r 2 N s ) 8 s 2 S; x 2 Omega : 14) December 11, 1997 DRAFT IEEE TRANSACTIONS ON PAMI, VOL. XX, NO. Y, MONTH 1999 ....

....the local clique set C s = fC 2 C; s 2 Cg. We provide two proofs for Proposition 1. The first method relies on the Mobius inversion formula (4) and follows Grimmett s [25] and Moussouris s [38] construction for the N potential V , Appendix A. The second proof is based on the ANOVA construction [18] for testing variable independence in a distribution, Appendix B. The proofs show that the strong MRF model is equivalent to the ANOVA construction. Even though equation (22) represents the general clique decomposition for P (x s ; x r ; r 2 N s ) it is subject to condition (43) Bishop et al. ....

[Article contains additional citation context not shown here]

S. E. Fienberg, The Analysis of Cross-Classified Categorical Data, vol. Second Edition. MIT Press, 1981.

....MRF which defines a site as being conditionally independent upon its non neighbouring sites given all of the neighbouring sites. This strong MRF model is equivalent to the Analysis of variance December 18, 1997 DRAFT 14TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION 4 (ANOVA) construction [7], which allows us to use the theorems from the ANOVA construction to estimate the LCPDF for the strong MRF model. We proved in [13] that given a neighbourhood system N , the LCPDF of a strong MRF is P (x s jx r ; r 2 N s ) Y C2Cs P (x s jx r ; r 2 C) n CsC ; 5) where n CsC = Gamma1) ....

S. E. Fienberg, The Analysis of Cross-Classified Categorical Data, vol. Second Edition. MIT Press, 1981.

No context found.

S. E. Fienberg, The Analysis of Cross--Classified Categorical Data, vol. Second Edition. MIT Press, 1981.

No context found.

S. E. Fienberg, The Analysis of Cross--Classified Categorical Data. MIT Press, 1981, vol. Second Edition.

No context found.

Fienberg SE. The Analysis of Cross-Classified Categorical Data. 2nd ed. Cambridge, MA: MIT Press; 1980

No context found.

S. E. Fienberg, The Analysis of Cross--Classified Categorical Data. MIT Press, 1981, vol. Second Edition.

No context found.

S. Fienberg. The Analysis of Cross-Classified Categorical Data. MIT Press, Cambridge, Mass., 2. edition, 1980.

No context found.

S. Fienberg. The Analysis of Cross-Classified Categorical Data. MIT Press, Cambridge, Mass., 2. edition, 1980.

No context found.

S. E. Fienberg, The Analysis of Cross--Classified Categorical Data, vol. Second Edition. MIT Press, 1981.

No context found.

Fienberg S.E. (1980), The analysis of cross-classified categorical data, MIT Press, Cambridge, Massachusetts.

No context found.

Stephen Fienberg. The Analysis of Cross-Classified Categorical Data, Second Edition. Cambridge, Mass., The MIT press, 1990

No context found.

Fienberg, S. E. (1980). The Analysis of Cross-Classified Categorical Data. The MIT Press, Cambridge, MA, second edition edition.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC