A Brief Introduction to Boosting

By Robert E. Schapire. 1999

How researchers see this paper

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

Citations in the corpus

Model Fusion via Optimal Transport (2019, methodology): “Ensemble methods [5–7] have long been in use in deep learning and machine learning in general.”
Machine Unlearning (2019, background): “Ensemble learning ap-3 Observe that this strategy holds even when the entire dataset D is replaced by the dataset for a particular country D c . proaches utilize boosting algorithms [46], even for ensembles of neural networks [47], to enhance accuracy.”
Trainable Sentence Planning for Complex Information Presentations in Spoken Dialog Systems (2004, background): “The SPR’s rules for ranking sp-trees are learned from a labeled set of sentence-plan training examples us-ing the RankBoost algorithm (Schapire, 1999).”
User re-authentication via mouse movements (2004, methodology): “Although techniques such as boosting [11, 33] or support vector machines [2] might obtain slightly higher classification accuracy, they require more computation during classification and further they obscure the decision making process.”
Discriminative Orderlet Mining for Real-Time Recognition of Human-Object Interaction (2014, methodology): “AdaBoosting [6] is employed here due to its good performance on feature selection and combination.”
Recognizing facial expression: machine learning and application to spontaneous behavior (2005, unknown): “At each step, the chosen filter can be shown to be uncorrelated with the output of the previous filters [6, 15].”
Boosting for transfer learning with multiple sources (2010, background): “Performance analysis The convergence properties of MultiSourceTrAdaBoost can be inherited directly from TrAdaBoost [4], whereas for TaskTrAdaBoost they can be inherited directly from AdaBoost [11, 18].”
Open problems in the security of learning (2008, background): “A popular learning technique known as boosting [26] exemplifies how a set of orthogonal learners could be constructed.”
Boosting for transfer learning (2007, result): “The conclusion given by Theorem 4 is the same as what has been shown in (Schapire, 1999).”
Beat the Machine: Challenging Humans to Find a Predictive Model's “Unknown Unknowns” (2015, methodology): “On the other hand, techniques like boosting [Schapire 1999] have a tremendous advantage by overweighting cases where the current model is incorrect.”
Detection of malicious code by applying machine learning classifiers on static features: A state-of-the-art survey (2009, methodology): “Adaboost was introduced by Freund and Schapire (1999) and among its many variants is the Adaboost.”
A trainable generator for recommendations in multimodal dialog (2003, methodology): “The sentence plan ranker (SPR) uses a set of rules for ranking sentence plans that were learned from a labelled set of sentence-plan training examples using the RankBoost algorithm, a type of boosting algorithm [8].”
An Introduction to Boosting and Leveraging (2002, methodology): “The present review differs from other reviews, such as the ones of [72, 165, 166], mainly in the choice of the presented material: we place more emphasis on robust algorithms for Boosting, on connections to other margin based approaches (such as support vector machines), and on connections to the Th”
Combining Independent Modules to Solve Multiple-choice Synonym and Analogy Problems (2003, methodology): “Boosting (Schapire 1999) also uses a logistic-regression-like rule to combine outputs of simple modules to perform state-of-the-art classiﬁcation.”
Fake News Detection Using Machine Learning Ensemble Methods (2020, background): “In the successive rounds, the weighted coeﬃcients are decreased for data points that are correctly classiﬁed and are increased for data points that are misclassiﬁed [35].”
The Greedy Miser: Learning under Test-time Budgets (2012, background): “Most prominently, Viola & Jones (2002) greedily train a cascade of weak clas-siﬁers with Adaboost (Schapire, 1999) for visual object recognition.”
SMOTE for Learning from Imbalanced Data: Progress and Challenges, Marking the 15-year Anniversary (2018, methodology): “These learning algorithms are iterative and they apply different weights to the data instances dynamically as the procedure evolves (Schapire, 1999).”
Multiple classifiers applied to multisource remote sensing data (2002, unknown): “For each x 2 S { if C (x ) = ! then weight (x ) = weight (x ) } 8.”