Results 1 -
3 of
3
Gini-Support Vector Machine: Quadratic Entropy Based Multi-class Probability Regression
- J. Machine Learning Research
, 2007
"... Many classification tasks require estimation of output class probabilities for use as confidence scores or for inference integrated with other models. Probability estimates derived from large margin classifiers such as support vector machines (SVMs) are often unreliable. We extend SVM large margin c ..."
Abstract
-
Cited by 2 (2 self)
- Add to MetaCart
Many classification tasks require estimation of output class probabilities for use as confidence scores or for inference integrated with other models. Probability estimates derived from large margin classifiers such as support vector machines (SVMs) are often unreliable. We extend SVM large margin classification to GiniSVM maximum entropy multi-class probability regression. GiniSVM combines a quadratic (Gini-Simpson) entropy based agnostic model with a kernel based similarity model. A form of Huber loss in the GiniSVM primal formulation elucidates a connection to robust estimation, further corroborated by the impulsive noise filtering property of the reverse water-filling procedure to arrive at normalized classification margins. The GiniSVM normalized classification margins directly provide estimates of class conditional probabilities, approximating kernel logistic regression (KLR) but at reduced computational cost. As with other SVMs, GiniSVM produces a sparse kernel expansion and is trained by solving a quadratic program under linear constraints. GiniSVM training is efficiently implemented by sequential minimum optimization or by growth transformation on probability functions. Results on synthetic and benchmark data, including speaker verification and face detection data, show improved classification performance and increased tolerance to imprecision over soft-margin SVM and KLR.
An Embedded Pedestrian Classifier for Automotive Applications
"... ABSTRACT: A hardware-friendly version of the well-known Support Vector Machine (SVM) is presented in this work, useful to implement its feed-forward phase on resource-limited architectures, such as Field Programmable Gate Arrays (FPGAs) or microcontrollers. In many embedded applications, a floating- ..."
Abstract
- Add to MetaCart
ABSTRACT: A hardware-friendly version of the well-known Support Vector Machine (SVM) is presented in this work, useful to implement its feed-forward phase on resource-limited architectures, such as Field Programmable Gate Arrays (FPGAs) or microcontrollers. In many embedded applications, a floating-point unit is seldom available, forcing a hardware-oriented reformulation. The system is tested on a pedestrian classification benchmarking dataset, showing how this proposal can be effectively applied in an automotive application. A modified version of the dataset is used in order to test the reliability to possible hardware artifacts.
