A Comparative Study of Cost-Sensitive Boosting Algorithms

By Kai Ming Ting. International Conference on Machine Learning, 2000

303 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/20212c96cee7b364f2d56aafe53d1d3be5377886

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

Cost-Sensitive Boosting (2011, methodology): “For example, CSB2 [25] modifies the weight update rule of AdaBoost to”
Feature Space Augmentation for Long-Tailed Data (2020, background): “It is typically accomplished by: 1) over-sampling or undersampling [37, 4, 50, 14, 22] to achieve an even data distribution across various classes, and/or 2) assigning proper weights to the loss terms corresponding to the tail classes [39, 20, 23, 27, 9, 48, 35, 11, 45].”
The Devil is in Classification: A Simple Framework for Long-tail Object Detection and Instance Segmentation (2020, background): “Loss Re-weighting [11], [19], [26], [38], [39] This line of works alleviate the bias by applying different weights to different samples or categories, such that tail classes or samples receive higher attention during training, thus improving the classification performance.”
Boosting for Learning Multiple Classes with Imbalanced Class Distribution (2006, methodology): “Within the bi-class applications, some variants of the AdaBoost algorithm in tackling the imbalance problem are reported, such as AdaCost [4], CSB1 and CSB2 [21], RareBoost [11] and AdaC1, AdaC2 and AdaC3 [20].”
Learning from Imbalanced Data (2009, methodology): “An empirical comparison over four imbalanced data sets of AdaC1, AdaC2, AdaC3, and AdaCost and two other similar algorithms, CSB1 and CSB2 [60], was performed in [58] using decision trees and a rule association system as the base classifiers.”
Learning to Segment the Tail (2020, background): “The former aims to re-balance the training samples across classes [16, 3, 11, 6]; while the latter focuses on assigning different weights to adjust the loss function [18, 40, 47, 7].”
An insight into classification with imbalanced data: Empirical results and current trends on using data intrinsic characteristics (2013, methodology): “Following this idea, many approaches have been developed by modifying the standard boosting weight-update mechanism in order to improve the performance on the minority class and the small disjuncts [30,44,61,69,74,112,117,122].”
Cost-sensitive boosting for classification of imbalanced data (2007, methodology): “In this paper, our study concentrates on AdaBoost [26,27], AdaCost [56], CSB2 [57], and the proposed three boosting algorithms, AdaC1, AdaC2, and AdaC3.”
A Review on Ensembles for the Class Imbalance Problem: Bagging-, Boosting-, and Hybrid-Based Approaches (2012, background): “These proposals usually differ in the way that they modify the weight update rule, among this family AdaCost [48], CSB1, CSB2 [49], RareBoost [97], AdaC1,”
Sensor-Based Abnormal Human-Activity Detection (2008, background): “Cost-sensitive learning addresses the issue of classification in the presence of different misclassification costs that are associated with different types of errors [7], [25] and are useful for learning problems where the classes are extremely unbalanced.”
Balanced Knowledge Distillation for Long-tailed Learning (2021, background): “Cost-sensitive re-weighting [10, 47, 23] assigns different weights to the classification loss terms corresponding to different classes [7, 23, 5] or different samples [29, 22, 27].”
Unequal-Training for Deep Face Recognition With Long-Tailed Noisy Data (2019, methodology): “Due to this complicated situation of long-tailed noisy face dataset, traditional methods, e.g . re-sampling [3] and cost-sensitive weighting [17], are no longer feasible.”
Learning a Unified Classifier Incrementally via Rebalancing (2019, background): “Previous efforts to tackle the class imbalance can be roughly divided into two groups: data resampling [13, 6, 12] and cost-sensitive learning [32, 17, 38, 16, 9].”
Learning from imbalanced data sets with boosting and data generation: the DataBoost-IM approach (2004, unknown): “5 decision tree [20], AdaBoostM1[8, 9], DataBoost [13], AdaCost [21], CSB2 [22] and SMOTEBoost [7] boosting algorithms.”
Data Mining for Imbalanced Datasets: An Overview (2005, methodology): “CSB ( Ting, 2000 ) and AdaCost boosting algorithms (Fan et al., 1999) update the weights of examples according to the misclassification costs.”
Small Area Estimation of the Homeless in Los Angeles: An Application of Cost-Sensitive stochastic Gradient Boosting (2010, methodology): “With L1(α) stochastic gradient boosting, our experience − both in this case study 7In a follow-up study of AdaCost and other cost-sensitive variations of AdaBoost, Ting (2000) shows that AdaCost stumbles in certain situations, and that this could be due to the algorithm’s weighting structure.”