On Combining Classifiers

By Josef Kittler, Mohamad Hatef, Robert P. W. Duin, Jiri Matas. 1998

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

Robust multimodal face and fingerprint fusion in the presence of spoofing attacks (2016, background): “'s classical framework [16]) in the presence of spoofing attacks is a further relevant sub-problem and addressed in this work.”
A Multilevel Mixture-of-Experts Framework for Pedestrian Classification (2011, methodology): “Altogether, there are six classifiers to be combined, using the sum and product of the individual decision values [ 26 ], as well as a fuzzy integration using Sugeno integrals [40].”
Design of effective neural network ensembles for image classification purposes (2001, background): “Recently, in the field of pattern recognition, the combination of an ensemble of neural networks has been proposed as an approach to the development of high performance image classification systems [1,4,5,7,8,11,29].”
EnsembleMatrix: interactive visualization to support machine learning with multiple classifiers (2009, unknown): “Simple rules such as majority vote, sum, product, maximum and minimum of the classifier outputs have been popular and often produce results better than individual classification system [21].”
Regularization studies of linear discriminant analysis in small sample size scenarios with application to face recognition (2005, background): “More sophisticated rules to combine the results from multiple classifiers can be found in (Kittler et al., 1998), but such a development is beyond the scope of this letter.”
Multimodal biometrics: An overview (2004, background): “General strategies for combining multiple classifiers have been suggested in [23] and [24].”
Ensemble based systems in decision making (2006, background): “Does the no-free-lunch theorem hold for combination rules as well? Many studies have been conducted to answer this question, including [18], [27], [28], [31], [36], [37], [42], [71], [72], among others.”
Machine Learning in Financial Crisis Prediction: A Survey (2012, background): “proposed to improve the classification performance of single classifiers [67].”
Query-adaptive late fusion for image search and person re-identification (2015, methodology): “First, previous works in biometric multi-modality fusion [16, 1] demonstrate that the product rule has very similar, if not superior, performance to the sum rule.”
A Theoretical Study on Six Classifier Fusion Strategies (2002, unknown): “Simple fusion methods are the most obvious choice when constructing a multiple classifier system [4], [5], [11], [13], [14], i.e., the support for class ! i , d i ðxÞ, yielded by the team is d i ðxÞ ¼ Fðd 1;i ðxÞ; . . . ; d L;i ðxÞÞ; i ¼ 1; 2; ð1Þ where F is the chosen fusion method.”
Information fusion in biometrics (2003, methodology): “These include majority voting , sum and product rules, k-NN classifiers, SVMs, decision trees, Bayesian methods, etc. (see for example Ben-Yacoub et al., 1999; Bigun et al., 1997; Dieckmann et al., 1997; Jain et al., 1999d; Kittler et al., 1998; Verlinde and Cholet, 1999).”
Deep ensemble learning of sparse regression models for brain disease diagnosis (2017, background): “85 The rationale of using a CNN is two aspects: (1) In machine learning, it is generally known that combining multiple classifiers is helpful for the improvement of the performance of individual classifiers (Kittler et al., 1998).”
'Online recognition of Chinese characters: the state-of-the-art (2004, background): “The parallel combination of multiple complementary classifiers can improve the classification accuracy [33], [54] and its promise in OLCCR has been justified in [119], [129].”
Multi-domain learning by confidence-weighted parameter combination (2010, methodology): “Kittler et al. (1998) provide a theoretical framework for combining classifiers.”
Ensembles of Deep LSTM Learners for Activity Recognition using Wearables (2017, methodology): “Independent of the cause for diversity in base classifiers (in our case LSTMs) the meta-optimisation objective for classifier Ensembles is to generate classifiers that are diverse and minimally correlated [Kittler et al. 1998].”
Breast cancer histopathological image classification using Convolutional Neural Networks (2016, methodology): “Product and Max rules (see [30] for details).”
Face Recognition from Long-Term Observations (2002, background): “Kittler et al in [9] present a statistical interpretation of a number of common methods for cross- modal fusion, such as the product, maximum , and majority rules, which are also appropriate for late integration over a set of observations from a single modality.”