Zheng, Z. and Webb, G. "Stochastic attribute selection committees with multiple boosting: Learning more accurate and more stable classifier committees." Proceedings of Pacific Asia International Conference on Knowledge Discovery and Data Mining (PAKDD-99), pp. 123-132, 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....2. The scoring algorithm in SBA 13 5. Empirical Evaluation We now compare the proposed technique with the state of the art classification system C4.5 (release 8) a Naive Bayesian (NB) classifier and boosted C4.5 [13, 34] Our version of boosted C4.5 is implemented and provided by Z. Zheng [39, 40]. We used 20 datasets in our experiments. Five (5) out of the 20 are our real life application datasets. The rest (12) of them are obtained from UCI Machine Learning Repository [28] We could not use many datasets in UCI Repository in our evaluation because in these datasets the class ....

Zheng, Z. and Webb, G. "Stochastic attribute selection committees with multiple boosting: Learning more accurate and more stable classifier committees." Proceedings of Pacific Asia International Conference on Knowledge Discovery and Data Mining (PAKDD-99), pp. 123-132, 1999.

....5 Experiments We now compare the classifiers built by msCBA, CBA, C4.5 (tree and rules, Release 8) RIPPER, NB, LB, and various combinations of msCBA, C4.5 and NB. The evaluations are done on 34 datasets from UCI ML Repository [9] We also used Boosted C4.5 (the code is obtained from Zijian Zheng [11]) in our comparison. We ran all the systems using their default settings. We could not compare with existing classifier combination methods as we were unable to obtain the systems. In all the experiments with msCBA, minconf is set to 50 . For t minsup, from our experience, once t minsup is ....

Zheng, Z. and Webb, G. 1999. Stochastic attribute selection committees with multiple boosting: Learning more accurate and more stable classifier committees. PAKDD-99.

....(Kohavi, 1995) were carried out for each algorithm. Some of these 37 domains contain missing attribute values, but some not. To simulate the situation where unseen examples contain certain amount of missing attribute values, we randomly introduce missing attribute values into test 3 In Zheng Webb (1998a; 1998b) We use 40 domains. To reduce the computational requirements of the experiments for this study, we exclude three largest domains from the test suite. The partial results in these 3 domains that we have got are consistent with our claims in this paper. 0 10 20 30 40 50 missing value level L ( in test ....

Zheng, Z. & Webb, G.I. 1998b. Stochastic Attribute Selection Committees with Multiple Boosting: Learning More Accurate and More Stable Classifier Committees. Technical Report (TR C98/13), School of Computing and Mathematics, Deakin University (available at http://www3.cm.deakin.edu.au/~zijian/Papers/sascmb-tr-C98-13.ps.gz).

....no study has been carried out on the effect of missing attribute values on the accuracy performance of committee learning techniques. In this paper, we study the robustness of four recently developed committee learning techniques including Boosting [4, 7] Bagging [3] Sasc [8] and SascMB [9] relative to C4.5 for tolerating missing values in test data. The motivation for this research is as follows. These four committee learning techniques can dramatically reduce the error of C4.5. This observation was obtained from experiments in domains containing no or a small amount of missing ....

....sample of the training data using a procedure that combines Boosting and Sasc by both stochasti2 cally selecting attribute subsets and adaptively modifying the distribution of the training set. Brief descriptions of Boost, Bag, and Sasc as well as a full description of SascMB can be found from [9] in the same proceedings. Readers may refer to the references mentioned above for further details about these algorithms. 3 Effects of Missing Values on Accuracy In this section, we use experiments to explore how the accuracy of the committee learning methods changes relative to that of C4.5, as ....

[Article contains additional citation context not shown here]

Zheng, Z. and Webb, G.I.: Stochastic attribute selection committees with multiple boosting: Learning more accurate and more stable classifier committees. Proceedings of the 3rd Pacific-Asia Conference on Knowledge Discovery and Data Mining. Berlin: Springer-Verlag (1999).

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