Collins, M., Schapire, R. E., and Singer, Y. (2002), "Logistic regression, Adaboost and Bregman distances," Machine Learning, 48:253--285. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
The Latent Maximum Entropy Principle - Wang, Schuurmans, Zhao (2002) (4 citations) (Correct)
....a maximum is achieved, otherwise monotonic convergence property can be violated for the sequential updates he proposes. In our case, EM IS uses a parallel update that avoids this di#culty. A sequential algorithm that maintains the monotonic convergence property can also be adapted as described in [11]. To compare EM IS to standard Boltzmann machine estimation techniques, first consider the derivation of a direct EM approach. In standard EM, given the previous parameters # , one solves for new parameters # by maximizing the auxiliary Q function with respect to # Q(#, # # ) p# # ....
M. Collins, R. Schapire and Y. Singer, "Logistic regression, AdaBoost and Bregman distances," Machine Learning, Vol. 48, No. 1-3, pp. 253-285, 2002
Constructing Boosting Algorithms from SVMs: an.. - Rätsch, Mika.. (2002) (8 citations) (Correct)
.... updates only the weight of the last hypothesis selected, while minimizing some exponential cost function [32] 47] 48] 39] It has been shown that this relates to coordi nate descent methods [49] 50] 35] barrier optimization techniques [8] 35] and to the Bregman algorithm [51] 52] [53], 35] In the next section we consider as an example two leveraging approaches for each of these two categories and adapt them to the one class problem. However, it will turn out that both of these approaches have certain drawbacks. Finally, in Section IV B we will develop our new method which ....
....the fol lowing simplified barrier objective: E(p,w) vp 3Elog(l exp[5l) t3 , 18) which does not contain the variables anymore. We therefore have only J 1 variables left to optimize. Furthermore, this functional form is very similar to the loss used in logistic regression (cf. 11] 47] [53]; in our case it has just the additional offset p and the scaling factor ) Usually, in a barrier algorithm one would optimize all J 1 parameters directly until the desired precision is reached 7This becomes clear by (12) and Section IV A. cf. Proposition 3) But this requires to know all ....
[Article contains additional citation context not shown here]
M. Collins, R.E. Schapire, and Y. Singer, "Logistic Regression, Adaboost and Bregman distances," in Proc. COLT, San Francisco, 2000, pp. 158 169, Morgan Kaufmann.
Convexity, Classification, and Risk Bounds - Peter Bartlett Bartlett (Correct)
No context found.
Collins, M., Schapire, R. E., and Singer, Y. (2002), "Logistic regression, Adaboost and Bregman distances," Machine Learning, 48:253--285.
Convexity, Classification, and Risk Bounds - Peter Bartlett Bartlett (Correct)
No context found.
Collins, M., Schapire, R. E., and Singer, Y. (2002), "Logistic regression, Adaboost and Bregman distances," Machine Learning, 48:253--285.
Geometric Context from a Single Image - Hoiem, Efros, Hebert (Correct)
No context found.
M. Collins, R. Schapire, and Y. Singer, "Logistic regression, adaboost and bregman distances," Machine Learning, vol. 48, no. 1-3, 2002.
Wishart Processes: A Statistical View of Reproducing.. - Zhang, Yeung, Kwok (2004) (Correct)
No context found.
M. Collins, R. E. Schapire, and Y. Singer, "Logistic regression, AdaBoost and Bregman distances," Machine Learning, vol. 47, no. 2/3, pp. 253--285, 2002.
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