Boosting and Other Ensemble Methods
Abstract
We compare the performance of three types of neural network-based ensemble techniques to that of a single neural network. The ensemble algorithms are two versions of boosting and committees of neural networks trained independently. For each of the four algorithms, we experimentally determine the test and training error curves in an optical character recognition (OCR) problem as both a function of training set size and computational cost using three architectures. We show that a single machine is best for small training set size while for large training set size some version of boosting is best. However, for a given computational cost, boosting is always best. Furthermore, we show a surprising result for the original boosting algorithm: namely, that as the training set size increases, the training error decreases until it asymptotes to the test error rate. This has potential implications in the search for better training algorithms.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Ensembling multiple CNN models is a common deep learning approach for improving prediction results.
“Ensembling multiple CNN models is a common deep learning approach for improving prediction results [27,28,29].”
From Fully Automated Placental Volume Quantification From<scp>3D</scp>Ultrasound for Prediction of Small‐for‐Gestational‐Age Infants (2022) · cites this paper as [28]
Cites
The Strength of Weak Learnability
Schapire, R. E. (1990) · Machine Learning · doi:10.1007/bf00116037 · Cited by 4,600 · see citing works
Hansen, L. K., & Salamon, P. (1990) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/34.58871 · Cited by 4,260 · see citing works
(1989) · neural information processing systems · Cited by 3,736 · see citing works
The Cascade-Correlation learning architecture
Fahlman, S. E., & Lebiere, C. J. (2018) · doi:10.1184/r1/6610403 · Cited by 2,391 · see citing works
Estimation of Dependences Based on Empirical Data
Vapnik, V. N. (2006) · Information science and statistics · doi:10.1007/0-387-34239-7 · Cited by 2,252 · see citing works
Self-citationSeung, H., Opper, M., & Sompolinsky, H. (1992) · Annual Conference Computational Learning Theory · doi:10.1145/130385.130417 · Cited by 2,193 · see citing works
When networks disagree: Ensemble methods for hybrid neural networks
Perrone, M., & Cooper, L. N. (1995) · World Scientific series in 20th century physics · doi:10.1142/9789812795885_0025 · Cited by 770 · see citing works
Cryptographic Limitations on Learning Boolean Formulae and Finite Automata
Kearns, M. J., & Valiant, L. G. (1994) · Journal of the ACM · doi:10.1145/174644.174647 · Cited by 739 · see citing works
Learning in feedforward layered networks: the tiling algorithm
Mézard, M., & Nadal, J. (1989) · Journal of Physics A Mathematical and General · doi:10.1088/0305-4470/22/12/019 · Cited by 433 · see citing works
Computer recognition of unconstrained handwritten numerals
Suen, C. Y., Nadal, C., Legault, R., et al. (1992) · Proceedings of the IEEE · doi:10.1109/5.156477 · Cited by 410 · see citing works
Consensus theoretic classification methods
Benediktsson, J. A., & Swain, P. H. (1992) · IEEE Transactions on Systems Man and Cybernetics · doi:10.1109/21.156582 · Cited by 357 · see citing works
Crytographic limitations on learning Boolean formulae and finite automata
Kearns, M. J., & Valiant, L. G. (1989) · ACM Symposium on Theory of Computing (STOC) · doi:10.1145/73007.73049 · Cited by 272 · see citing works
Training Connectionist Networks with Queries and Selective Sampling
Atlas, L., Cohn, D. A., & Ladner, R. E. (1989) · neural information processing systems · Cited by 220 · see citing works
BOOSTING PERFORMANCE IN NEURAL NETWORKS
Drucker, H., Schapire, R. E., & Simard, P. (1993) · International Journal of Pattern Recognition and Artificial Intelligence · doi:10.1142/s0218001493000352 · Cited by 185 · see citing works
Self-citationImproving Performance in Neural Networks Using a Boosting Algorithm
Drucker, H., Schapire, R. E., & Simard, P. (1992) · neural information processing systems · Cited by 157 · see citing works
Self-citationHierarchies of adaptive experts
Jordan, M. I., & Jacobs, R. A. (1991) · Cited by 152 · see citing works
Boosting a weak learning algorithm by majority
Freund, Y. (1990) · Conference on Learning Theory · doi:10.5555/92571.92640 · Cited by 130 · see citing works
Document image defect models and their uses
Baird, H. S. (2002) · Proceedings of the International Conference on Document Analysis and Recognition · doi:10.1109/icdar.1993.395781 · Cited by 83 · see citing works
Constructing Hidden Units using Examples and Queries
Baum, E. B., & Lang, K. (1990) · neural information processing systems · Cited by 73 · see citing works
High-performance reading machines
Srihari, S. N. (1992) · Proceedings of the IEEE · doi:10.1109/5.156474 · Cited by 56 · see citing works
Computer aided cleaning of large databases for character recognition
Matic, N., Guyon, I., Bottou, L., et al. (2003) · IAPR International Conference on Pattern Recognition (ICPR) · doi:10.1109/icpr.1992.201784 · Cited by 36 · see citing works
Generalization Abilities of Cascade Network Architecture
Littmann, E., & Ritter, H. J. (1992) · neural information processing systems · Cited by 27 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Measures of Diversity in Classifier Ensembles and Their Relationship with the Ensemble Accuracy
Kuncheva, L. I., & Whitaker, C. J. (2003) · Machine Learning · doi:10.1023/a:1022859003006 · Cited by 2,335 · see citing works
Digital medicine and the curse of dimensionality
Berisha, V., Krantsevich, C., Hahn, P. R., Hahn, S., Dasarathy, G., Turaga, P., & Liss, J. (2021) · npj Digital Medicine · doi:10.1038/s41746-021-00521-5 · Cited by 344 · see citing works
Probabilistic Incremental Program Evolution
Salustowicz, R., & Schmidhuber, J. (1997) · Evolutionary Computation · doi:10.1162/evco.1997.5.2.123 · Cited by 249 · see citing works
Ensemble Deep Learning for Biomedical Time Series Classification
Jin, L.-P., & Dong, J. (2016) · Computational Intelligence and Neuroscience · doi:10.1155/2016/6212684 · Cited by 60 · see citing works
Schwartz, N., Oguz, I., Wang, J., Pouch, A., Yushkevich, N., Parameshwaran, S., Gee, J., Yushkevich, P., & Oguz, B. (2022) · Journal of Ultrasound in Medicine · doi:10.1002/jum.15835 · Cited by 11 · see citing works
- Authors
- Drucker, H., Cortes, C., Jackel, L. D., LeCun, Y., & Vapnik, V.
- Affiliations
- AT&T (United States) (US)
- Venue
- Neural Computation
- Publisher
- MIT Press - Journals
- Published
- 1994-11
- Volume
- 6
- Issue
- 6
- Pages
- 1289-1301
- ISSN
- 0899-7667 (ISSN-L)
- 1530-888X
- OpenAlex
- W2084362125
- Cited by
- 357
- References
- 22
- Citations by year
- 2026: 3 · 2025: 8 · 2024: 10 · 2023: 17 · 2022: 15 · 2021: 19 · 2020: 16 · 2019: 14 · 2018: 17 · 2017: 9 · 2016: 13 · 2015: 9 · 2014: 7 · 2013: 14 · 2012: 13
- FWCI
- 7.72
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Neural Computation (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Algorithms
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Keywords
- boosting, neural network ensembles, committee machine, optical character recognition, training data size, test error, training error, computational cost
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
