Do we need hundreds of classifiers to solve real world classification problems?

By M. Delgado, Eva Cernadas, Senén Barro, Dinani Gomes Amorim. 2014

3,241 citations recorded by the source corpus.

How researchers see this paper

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

In Information Systems

Citations in the corpus

Mathematical optimization in classification and regression trees (2021, background): “Instead, one can offer to individuals counterfactual explanations (Fernández et al. 2020; Lucic et al. 2020; Mothilal et al. 2020).”
Self-Normalizing Neural Networks (2017, methodology): “[10] for data preparation and defining training and test sets.”
Optimal randomized classification trees (2021, background): “[36], but having a complex and hardly interpretable decision structure.”
Monotonic Calibrated Interpolated Look-Up Tables (2015, methodology): “In the ad-query scoring case study, we do include a comparison to boosted monotonic stumps that had been optimized by an expert in those models - it performs poorly (as do boosted stumps on average (Fernandez-Delgado et al., 2014)).”
Encrypted statistical machine learning: new privacy preserving methods (2015, methodology): “Unsurprisingly, the traditional random forest tends to perform best in the more challenging data sets (Fernández-Delgado et al., 2014), though only in 4 of the data sets is it clearly outperforming all the other methods by the AUC metric.”
Review of Deep Learning Algorithms and Architectures (2019, background): “Nonetheless [27] evaluated 179 classifiers and concluded that parallel random forest or parRF_t, which is essentially parallel implementation of variation of decision tree, performed the best.”
Neural Kernels Without Tangents (2020, methodology): “(2019)), CIFAR-100 (Krizhevsky (2009)) and 90 UCI datasets (Fernández-Delgado et al. (2014)).”
The Impact of Automated Parameter Optimization on Defect Prediction Models (2018, background): “[29], who find that random forest tends to yield the most frequently”
Multicenter Comparison of Machine Learning Methods and Conventional Regression for Predicting Clinical Deterioration on the Wards (2016, background): “For example, Fernandez-Delgado et al compared over 100 different techniques in 121 datasets, many of which were small and in nonmedical fields, and found that the random forest algorithm was the most accurate method (6).”
Fair lending needs explainable models for responsible recommendation (2018, background): “Can complex models be explained and proven to be fair on a case-by-case basis? desire for explainability appears to clash with the desire to use complex models that may provide improved or classification accuracy [25, 50]. kind of complexity has sometimes been termed “opacity at scale” [9].”
Mitigating Concept Drift via Rejection (2018, methodology): “Random forests: Random forests as introduced by Breiman [20] constitute one of the current state-of-the-art classifiers [21], offering a classification as an ensemble of decision trees.”
Sparsity in Optimal Randomized Classification Trees (2020, background): “Standard CARTs or Random Forests (RFs) [5, 7, 13, 16] cannot manage it due to the greedy construction of the trees.”
Video-Based Facial Micro-Expression Analysis: A Survey of Datasets, Features and Algorithms (2021, background): “Pattern classification/recognition has a long history [107], and many well-developed classifiers have already been proposed; the reader is referred to [108] for more details.”