Al-Ghoneim, K. and Vijaya Kumar, B. V. K. (1995). Learning ranks with neural networks (Invited paper). In Applications and Science of Artificial Neural Networks, Proceedings of the SPIE, volume 2492, pages 446--464.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... have been mathematically analyzed both for classification (Tumer and Ghosh, 1995c; Tumer and Ghosh, 1996) and regression problems (Perrone and Cooper, 1993a; Hashem and Schmeiser, 1993) Some researchers have investigated non linear combiners using rank based information (Ho et al. 1994; Al Ghoneim and Vijaya Kumar, 1995), belief based methods (Rogova, 1994; Yang and Singh, 1994; Xu et al. 1992) or voting schemes (Hansen and Salamon, 1990; Battiti and Colla, 1994) We have introduced order statistics combiners, and analyzed their properties (Tumer and Ghosh, 1995b; Tumer and Ghosh, 1995c) Wolpert introduced ....

....will also report the bagging results, i.e. when majority voting is used. 4.4 Weighted Experts The final method of correlation reduction that we present has a different flavor than the previous ones. The method is based on the mixture of experts framework (Haykin, 1994; Jacobs et al. 1991; Xu et al. 1995), where the output is a weighted sum of the outputs of individual networks or experts . The weights are determined by a gating network, and are a function of the inputs. In a given region of the input space, a particular expert will be weighted more than others. Moreover, the parameter updates ....

[Article contains additional citation context not shown here]

Al-Ghoneim, K. and Vijaya Kumar, B. V. K. (1995). Learning ranks with neural networks (Invited paper). In Applications and Science of Artificial Neural Networks, Proceedings of the SPIE, volume 2492, pages 446--464.

....with different methods of computing the proper classifier weights [6, 21, 24, 27] Such linear combining techniques have been mathematically analyzed for regression problems [21, 32] but not for classification. Some researchers have investigated non linear combiners using rank based information [1, 23], or voting schemes [4] Methods for combining 2 f ind f comb Classifier 1 Classifier N Classifier m Feature Set 2 Feature Set 1 Feature Set M Raw Data from Observed Phenomenon Combiner Figure 1: Combining Strategy. The solid lines leading to f ind represent the decision of a specific ....

K. Al-Ghoneim and B. V. K. Vijaya Kumar. Learning ranks with neural networks (Invited paper). In Applications and Science of Artificial Neural Networks, Proceedings of the SPIE, volume 2492, pages 446--464, April 1995. 30

.... or by taking a weighted average (e.g. Perrone and Cooper, 1993; Hashem and Schmeiser, 1993) Non linear combining methods: Other non linear combining methods that have been proposed include Dempster Shafer belief based methods, e.g. Rogova, 1994) combining using rank based information (e.g. Al Ghoneim and Kumar, 1995), voting (e.g. Hansen and Salamon, 1990) and order statistics (Tumer and Ghosh, 1995) Supra Bayesian: Jacobs (1995) contrasts supra Bayesian with linear combinations. The underlying philosophy of supra Bayesian approach is that 6 the opinions of the experts are themselves data. Therefore the ....

Al-Ghoneim, K. and Vijaya Kumar, B.V.K. (1995) Learning ranks with neural networks (Invited paper). In Applications and Science of Artificial Neural Networks, Proceedings of the SPIE, volume 2492, pg 446-464.

....on linear methods, nonlinear combiners have also been studied. For example, Ho and Hull study multiple classifier systems that provide rank information, and apply their results to word recognition [78] Al Ghoneim and Kumar discuss networks specifically designed to learn and use rank information [2]. Battiti and Colla discuss voting schemes in network groups, and relate the results to classifier confusion levels [17] Breiman shows that using majority voting can lead to improved classification in [25] Cho and Kim propose a non linear combining scheme based on fuzzy logic [30, 31] Cooper ....

K. Al-Ghoneim and B. V. K. Vijaya Kumar, Learning ranks with neural networks (Invited paper), in Applications and Science of Artificial Neural Networks, Proceedings of the SPIE, vol. 2492, April 1995, pp. 446-- 464.

....and have comparable success. However, such combiners are susceptible to outliers and to unevenly performing classifiers. In the second category, meta learners, i.e. either sets of combining rules, or full fledged classifiers acting on the outputs of the individual classifiers, are constructed [1], 9] 28] This type of combining is more general, but suffers from all the problems associated with the extra learning (e.g. overparameterizing, lengthy training time) Both these methods are in fact ill suited for problems where most (but not all) classifiers perform within a wellspecified ....

K. Al-Ghoneim and B. V. K. Vijaya Kumar. Learning ranks with neural networks (Invited paper). In Applications and Science of Artificial Neural Networks, Proceedings of the SPIE, volume 2492, pages 446--464, April 1995.

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