Abstract
We introduce Gaussian Margin Machines (GMMs), which maintain a Gaussian distribution over weight vectors for binary classification. The learning algorithm for these machines seeks the least informative distribution that will classify the training data correctly with high probability. One formulation can be expressed as a convex constrained optimization problem whose solution can be represented linearly in terms of training instances and their inner and outer products, supporting kernelization. The algorithm admits a natural PAC-Bayesian justification and is shown to minimize a quantity directly related to a PAC-Bayesian generalization bound. A preliminary evaluation on handwriting recognition data shows that our algorithm improves on SVMs for the same task, achieving lower test error and lower test error variance.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 105-112 |
| Number of pages | 8 |
| Journal | Journal of Machine Learning Research |
| Volume | 5 |
| State | Published - 2009 |
Fingerprint
ASJC Scopus subject areas
- Artificial Intelligence
- Software
- Control and Systems Engineering
- Statistics and Probability
Cite this
Gaussian margin machines. / Crammer, Koby; Mohri, Mehryar; Pereira, Fernando.
In: Journal of Machine Learning Research, Vol. 5, 2009, p. 105-112.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Gaussian margin machines
AU - Crammer, Koby
AU - Mohri, Mehryar
AU - Pereira, Fernando
PY - 2009
Y1 - 2009
N2 - We introduce Gaussian Margin Machines (GMMs), which maintain a Gaussian distribution over weight vectors for binary classification. The learning algorithm for these machines seeks the least informative distribution that will classify the training data correctly with high probability. One formulation can be expressed as a convex constrained optimization problem whose solution can be represented linearly in terms of training instances and their inner and outer products, supporting kernelization. The algorithm admits a natural PAC-Bayesian justification and is shown to minimize a quantity directly related to a PAC-Bayesian generalization bound. A preliminary evaluation on handwriting recognition data shows that our algorithm improves on SVMs for the same task, achieving lower test error and lower test error variance.
AB - We introduce Gaussian Margin Machines (GMMs), which maintain a Gaussian distribution over weight vectors for binary classification. The learning algorithm for these machines seeks the least informative distribution that will classify the training data correctly with high probability. One formulation can be expressed as a convex constrained optimization problem whose solution can be represented linearly in terms of training instances and their inner and outer products, supporting kernelization. The algorithm admits a natural PAC-Bayesian justification and is shown to minimize a quantity directly related to a PAC-Bayesian generalization bound. A preliminary evaluation on handwriting recognition data shows that our algorithm improves on SVMs for the same task, achieving lower test error and lower test error variance.
UR - http://www.scopus.com/inward/record.url?scp=84862273709&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84862273709&partnerID=8YFLogxK
M3 - Article
AN - SCOPUS:84862273709
VL - 5
SP - 105
EP - 112
JO - Journal of Machine Learning Research
JF - Journal of Machine Learning Research
SN - 1532-4435
ER -