Experiments with a New Boosting Algorithm

By Yoav Freund, Robert E. Schapire. 1996

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistical Learning Theory

Citations in the corpus

Random Forests (2001, methodology): “In its original version, Adaboost (Freund & Schapire, 1996) is a deterministic algorithm that selects the weights on the training set for input to the next classiﬁer based on the mis-classiﬁcations in the previous classiﬁers.”
An optimization approach to learning falling rule lists (2017, methodology): “0 [23], respectively, (6) a random forest [6] of decision trees trained with cost-sensitive CART, (7) a boosted tree classiﬁer using AdaBoost [10] on trees trained with cost-sensitive CART, and (8) a decision list using RIPPER k [7], and we computed the true positive rate and the false positive rate”
The Random Subspace Method for Constructing Decision Forests (1998, methodology): “Two training set subsampling methods have been proposed: bootstrapping [4] and boosting [6].”
A Survey on Multi-view Learning (2013, methodology): “AdaBoost (Freund and Schapire, 1996) is another well-known ensemble learning algorithm, in which the principal idea is to train a new model to compensate for the errors made by earlier models.”
PREDICTIVE LEARNING VIA RULE ENSEMBLES (2008, methodology): “AdaBoost [Freund and Schapire (1996)] uses exponential loss, L ( y, ˆ y ) = exp( y · ˆ y ) for y ∈ {− 1 , 1 , and is equivalent to setting ν = 1 and η = N in Algorithm 1.”
Kernel methods in machine learning (2007, neutral): “(76) This is related to the exponential loss used in the AdaBoost method of Freund and Schapire [53].”
Online Human Action Detection using Joint Classification-Regression Recurrent Neural Networks (2016, methodology): “The method in [29] uses a traditional boosting algorithm [31] and its score is signiﬁcantly lower than other methods.”
Nearly unbiased variable selection under minimax concave penalty (2010, methodology): “The LASSO is easy to compute [Osborne, Presnell and Turlach (2000a, 2000b) and Efron et al. (2004)] and has the interpretation as boosting [Schapire (1990), Freund and Schapire (1996) and Friedman, Hastie and Tibshirani (2000)].”
Node harvest: simple and interpretable regression and classication (2009, neutral): “Use of the L 2 -loss as a convex surrogate for misclassiﬁcation error is thus not only appropriate for NH, it is even beneﬁcial since it allows for an interpretation of the predictions ˆ Y ( x ) as weighted empirical node means.”
PENALIZED LINEAR UNBIASED SELECTION (2007, background): “…(1.2) is the LASSO (Tibshirani, 1996) with ρ(t;λ) = λ|t|, which is easy to compute (Osborne, Presnell and Turlach, 2000a, 2000b; Efron, Hastie, Johnstone and Tibshirani, 2004) and has the interpretation as boosting (Schapire, 1990; Freund and Schapire, 1996; Friedman, Hastie and Tibshirani, 2000).”
Boosting for high-dimensional linear models (2006, background): “Freund and Schapire’s [11] AdaBoost algorithm for classification has attracted much attention in the machine learning community (cf.”
Survey : Domain Adaptation Algorithms for Natural Language Processing (2012, unknown): “The TrAdaBoost algorithm is an extension to the original AdaBoost algorithm (Freund and Schapire, 1996), which is tailored for transfer learning purposes.”
Loss Functions for Binary Class Probability Estimation and Classification: Structure and Applications (2005, background): “(45) Now q = q(F ) is the inverse of the link function F = F (q), hence F ′ (q) = 1/q ′ (F (q)) and from (45) F ′ (q) = ω(q) = L ′ 1 (1 − q) + L ′ 0 (q) , (46) where the second equality follows from L ′ 1 (1 − q) + L ′ 0 (q) = (1 − q) ω(q) + q ω(q), which is the sum of Equations (9).”
Machine Learning: Algorithms, Real-World Applications and Research Directions (2021, unknown): “[35] and also known as “meta-learning”.”
Bagging, Boosting and the Random Subspace Method for Linear Classifiers (2002, methodology): “Boosting, proposed by Freund and Schapire [5], is another technique to combine weak classifiers having a poor performance in order to get a classification rule with a better performance.”