P. Chan, S. Stolfo, and D. Wolpert, editors. Working Notes for the AAAI-96 Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, Portland, OR, 1996. AAAI.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....learning is a key technique for datamining. Prediction accuracy and computational requirements are two primary concerns with this type of learning. In order to improve the prediction accuracy of classifiers, classifier committee 1 learning techniques have been developed with great success [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14]. This type of technique generates several classifiers to form a committee by using a single base learning algorithm. At the classification stage, the committee members vote to make the final decision. Bagging [6] and Boosting [15, 3, 4, 2, 12] as two representative methods of this type, can ....

.... trees by manually changing learning parameters [16] error correcting output codes [17] generating different classifiers by randomizing the base learning process [8, 9] and learning option trees [18] A collection of recent research in this area and reviews of related methods can be found in [10, 1, 9]. In this paper, we investigate an approach, namely MB (Multiple Boosting) 2 to generating committees, which is more accurate than Bagging and is more stable than Boosting. MB creates multiple subcommittees by incorporating Bagging into Boosting using the multiboosting technique [19] We ....

Chan, P., Stolfo, S., and Wolpert, D. (eds). Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (available at http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon, 1996.

....inside a statistical tagger. The point here is that the context is not restricted to the n Gamma1 preceding tags as in the n gram formulation. Instead, it is extended to all the contex tual information used for learning the decision trees. The Viterbi algorithm (described for instance in (DeRose, 1988)) in which n gram probabilities are substituted by the application of the corresponding decision trees, allows the calculation of the most likely sequence of tags with a linear cost on the sequence length. However, one problem appears when applying conditionings on the right context of the ....

In P. Chan, editor, Working Notes of the AAAI Workshop on Integrating Multiple Learned Models, pages 15--21. K. W. Church. 1988. A Stochastic Parts Program and Noun Phrase Parser for Unrestricted Text.

.... codes [Dietterich and Bakiri, 1995] generating different classifiers by randomizing the base learning process [Dietterich and Kong, 1995; Ali, 1996] and learning option trees [Kohavi and Kunz, 1997] A collection of recent research in this area and reviews of related methods can be found in [Chan et al. 1996; Dietterich, 1997; Ali, 1996] In this paper, we investigate an approach, namely MB (Multiple Boosting) 2 to generating committees, which is more accurate than Bagging and is more stable than Boosting. MB creates multiple subcommittees by incorporating Bagging into Boosting using the ....

Chan, P., Stolfo, S., and Wolpert, D. (eds): Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (available at http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon.

....requirements are two primary concerns. To increase the prediction accuracy of classifiers, classifier committee 1 learning techniques have been developed with great success (Freund 1996; Freund and Schapire 1996a; 1996b; Quinlan 1996; Breiman 1996a; 1996b; Dietterich and Kong 1995; Ali 1996; Chan, Stolfo, and Wolpert 1996; Schapire, Freund, Bartlett, and Lee 1997; Domingos 1997; Bauer and Kohavi 1998) especially Boosting 2 (Freund and Schapire 1996b; Quinlan 1996; Bauer and Kohavi 1998) This type of technique generates several classifiers to form a committee by using a single base learning algorithm. At the ....

.... generating different classifiers by randomizing the base learning process (Dietterich and Kong 1995; Ali 1996) which is similar to Sasc (Zheng and Webb 1998) Reviews of related methods are provided in Dietterich (1997) and Ali (1996) A collection of recent research in this area can be found from Chan et al. 1996). Base on the observation that both Boosting and Sasc can significantly increase the prediction accuracy of decision trees but through different mechanisms, we developed another technique to further improve the accuracy of decision trees (Zheng, Webb, and Ting 1998) The new approach is called ....

Chan, P., Stolfo, S., and Wolpert, D. 1996. Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (available at http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon.

....to parallel or distributed processing, while Boosting and SascB are not. This is an important characteristic for datamining in large datasets. 1 Introduction To increase the prediction accuracy of classifiers, classifier committee 1 learning techniques have been developed with great success [2, 3, 4, 5, 6, 7, 8, 9, 10, 11]. This type of technique generates several classifiers to form a committee by using a single base learning algorithm. At the classification stage, the committee members vote to make the final decision. Bagging [5] and Boosting [2, 3, 6, 10, 12] as two representative methods of this type, can ....

.... show that Sasc, like Boosting, can also significantly reduce the error rate of decision tree learning [13] In addition, Sasc is more stable than Boosting [13] Sasc [13] is a minor variant of a class of committee learning algorithm that learns a committee by randomizing the base learning process [1, 7, 8, 9]. While Sasc has not been directly compared with these alternatives, comparisons of reported results suggest that its performance is comparable to that of others. Base on the observation that both Boosting and Sasc can significantly increase the prediction accuracy of decision trees but through ....

Chan, P., Stolfo, S., and Wolpert, D. (eds): Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon (1996).

.... learning, decision tree learning, inductive learning, committee learning, datamining 1 Introduction Classifier committees have been the focus of much recent attention (Freund, 1996; Freund and Schapire, 1996a; 1996b; Quinlan, 1996; Breiman, 1996a; 1996b; Dietterich and Kong, 1995; Ali, 1996; Chan, Stolfo, and Wolpert, 1996; Ali and Pazzani, 1996; Schapire, Freund, Bartlett, and Lee, 1997; Domingos, 1997; Bauer and Kohavi, 1998) With this type of approach, a set of classifiers is generated using a single base learning algorithm to form a committee. The committee members vote to decide the final classification. ....

.... of conditions to FOIL (Ali and Pazzani, 1996) Finally, dif2 ferent base learning algorithms can be used for learning different classifiers in committees (Wolpert, 1992; Zhang, Mesirov, and Waltz, 1992) A collection of recent research in this area and reviews of related methods can be found in Chan, Stolfo, and Wolpert (1996), Dietterich (1997) and Ali (1996) In contrast to Bagging and Boosting, this paper studies an alternative approach to generating different classifiers to form a committee, namely SASC (Stochastic Attribute Selection Committees) SASC builds different classifiers by stochastically modifying the ....

Chan, P., Stolfo, S., and Wolpert, D. (eds): Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (available at http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon (1996).

....Moreover, like Bagging, Sasc is amenable to parallel and distributed processing while Boosting is not. 1 Introduction Classifier committees have been the focus of much recent attention (Freund 1996; Freund and Schapire 1996; Quinlan 1996; Breiman 1996a; 1996b; Dietterich and Kong 1995; Ali 1996; Chan, Stolfo, and Wolpert 1996; Ali and Pazzani 1996; Schapire, Freund, Bartlett, and Lee 1997; Bauer and Kohavi 1998) With this type of approach, a set of classifiers is generated using a single base learning algorithm to form a committee. The committee members vote to decide the final classification. Bagging (Breiman 1996a) ....

.... learning by adding random selection of conditions to FOIL (Ali and Pazzani 1996) Finally, different base learning algorithms can be used for learning different classifiers in committees (Wolpert 1992) A collection of recent research in this area and reviews of related methods can be found in Chan, Stolfo, and Wolpert (1996), Dietterich (1997) and Ali (1996) In contrast to Bagging and Boosting, this paper studies an alternative approach to generating different classifiers to form a committee, namely SASC (Stochastic Attribute Selection Committees) SASC builds different classifiers by stochastically modifying the ....

Chan, P., Stolfo, S., and Wolpert, D. (eds): Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (available at http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon (1996).

....Bagging and Sasc, SascBag is amenable to parallel and distributed processing while Boosting is not. This gives SascBag another advantage over Boosting for parallel machine learning and datamining. 1 Introduction Much recent research effort has been exerted to study classifier committee learning [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]. With this type of approach, a set of classifiers is generated using a single base learning algorithm to form a committee. The committee members vote to decide the final classification. Bagging [5] and Boosting [13, 2, 3, 1, 11] as two representative methods of this type, can significantly ....

.... committees for first order learning by adding random selection of conditions to FOIL [10] Finally, different base learning algorithms can be used for learning different classifiers in committees [23, 24] A collection of recent research in this area and reviews of related methods can be found in [9, 25, 8]. As an alternative approach to creating classifier committees, the stochastic attribute selection committee learning method can also significantly reduce the error rates of decision tree learning [7, 8, 16] It builds different classifiers by stochastically modifying the set of attributes ....

Chan, P., Stolfo, S., and Wolpert, D. (eds): Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (available at http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon (1996).

....Data mining. 1 Introduction In order to increase the prediction accuracy of classifiers, classifier committee 1 learning techniques have been developed with great success (Freund 1996; Freund and Schapire 1996a; 1996b; Quinlan 1996; Breiman 1996a; 1996b; Dietterich and Kong 1995; Ali 1996; Chan, Stolfo, and Wolpert 1996; Schapire, Freund, Bartlett, and Lee 1997; Domingos 1997; Bauer and Kohavi 1998; Zheng and Webb 1998) especially Boosting 2 (Freund and Schapire 1996b; Quinlan 1996; Bauer and Kohavi 1998) This type of technique generates several classifiers to form a committee by using a single base learning ....

.... generating different classifiers by randomizing the base learning process (Dietterich and Kong 1995; Ali 1996) which is similar to Sasc (Zheng and Webb 1998) Reviews of related methods are provided in Dietterich (1997) and Ali (1996) A collection of recent research in this area can be found from Chan et al. 1996). In the following section, we briefly describe the Boosting and Sasc techniques for decision tree learning. Section 3 presents the SascB method of combining Boosting and Sasc. SascB is, then, empirically evaluated using a representative collection of natural domains. Finally, we summarize our ....

Chan, P., Stolfo, S., and Wolpert, D. 1996. Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms (available at http://www.cs.fit.edu/~imlm/papers.html), Portland, Oregon.

.... [15] errorcorrecting output codes [8] and generating different classifiers by randomizing the base learning process [9, 1] which is similar to SASC [21] Reviews of related methods are provided by Dietterich [7] and Ali [1] A collection of recent research in this area can be found from [6]. In the following section, we briefly describe the Boosting and SASC techniques for decision tree learning. Section 3 presents the SASCB method of combining Boosting and SASC. SASCB is, then, empirically evaluated using a representative collection of natural domains. Finally, we summarize our ....

P. Chan, S. Stolfo, and D. Wolpert. Working Notes of AAAI Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, 1996. (available at http://www.cs.fit. edu/~imlm/papers.html), Portland, Oregon.

....collated pages in parallel. Using a merge sort like algorithm, ranked pages on each processor are recursively merged into the final sorted list of ranked pages. Using different learning algorithms and combining the learned models has been demonstrated to be effective in improving performance [3, 4, 8, 37, 46]. We plan to investigate using multiple estimators to analyze and combine the results in parallel. Meta learning [6] is one combining approach and we intend to examine other voting based methods [7] Properly evaluating how the user responds to the results returned by our search engine is ....

P. Chan, S. Stolfo, and D. Wolpert, editors. Working Notes for the AAAI-96 Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, Portland, OR, 1996. AAAI.

....performance of an ensemble computed over those that remain after pruning. 1. INTRODUCTION Recently, there has been considerable interest in metalearning techniques that combine or integrate an ensemble of models computed by the same or di#erent learning algorithms over multiple data subsets [7, 10]. An advantage # This research is supported by the Intrusion Detection Program (BAA9603) from DARPA (F30602 96 1 0311) NSF (IRI 96 32225 and CDA 96 25374) and NYSSTF (423115445) Supported in part by an IBM fellowship. Permission to make digital or hard copies of all or part of this work ....

P. Chan, S. Stolfo, and D. Wolpert, editors. Working Notes for the AAAI-96 Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, Portland, OR, 1996.

....to interface with the JAM system and can subsequently be transfered and executed at a different site, provided, of course, that both the receiving site and the native program are compatible. 5 Compatibility Combining multiple models has been receiving increased attention in the literature [19, 11]. In much of the prior work on combining multiple models, it is assumed that all models originate from different subsets (not necessarily distinct) of a single data set as a means to increase accuracy, e.g. by imposing probability distributions over the instances of the training set, or by ....

P. Chan, S. Stolfo, and D. Wolpert, editors. Working Notes for the AAAI-96 Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, Portland, OR, 1996.

.... and NYSSTF (423115 445) Supported in part by an IBM fellowship 1 Introduction Recently, there has been considerable interest in meta learning techniques that combine or integrate an ensemble of models computed by the same or di#erent learning algorithms over multiple data subsets [6, 8]. An advantage of such an approach is that it can improve both e#ciency and scalability by executing the machine learning processes in parallel and on (possible disjoint) subsets of the data (a data reduction technique) Moreover, it can produce a higher quality final classification model, also ....

P. Chan, S. Stolfo, and D. Wolpert, editors. Working Notes for the AAAI-96 Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, Portland, OR, 1996. AAAI.

....to interface with the JAM system and can subsequently be transfered and executed at a di#erent site, provided, of course, that both the receiving site and the native program are compatible. 5 Compatibility Combining multiple models has been receiving increased attention in the literature [17, 11]. In much of the prior work on combining multiple models, it is assumed that all models originate from di#erent subsets (not necessarily distinct) of a single data set as a means to increase accuracy, e.g. by imposing probability distributions over the instances of the training set, or by ....

P. Chan, S. Stolfo, and D. Wolpert, editors. Working Notes for the AAAI-96 Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, Portland, OR, 1996. AAAI.

....many realistic problems and databases. One means to address this problem is to apply various inductive learning programs over the distributed subsets of data in parallel and integrate the resulting classification models or classifiers in some principled fashion to boost overall predictive accuracy [8, 20]. This approach has two advantages, first it uses serial code (standard o# the shelf learning programs) at multiple sites without the time consuming process of writing parallel programs and second, the learning processes can use small subsets of data that can fit in main memory (a data reduction ....

P. Chan, S. Stolfo, and D. Wolpert, editors. Working Notes for the AAAI-96 Workshop on Integrating Multiple Learned Models for Improving and Scaling Machine Learning Algorithms, Portland, OR, 1996. AAAI.

No context found.

In Chan, P. (Ed.), Working Notes of the AAAI Workshop on Integrating Multiple Learned Models, pp. 15--21. Dietterich, T. G., & Bakiri, G. (1991). Errorcorrecting output codes: A general method for improving multiclass inductive learning programs.

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