Second-order training improves variable projection SVM for road sensor data
Second-order optimization of variable projection SVM models and road abnormality detection
Machine Learning
Summary
Training support vector machines (SVMs) can be slow and complicated, especially when advanced techniques like variable projection are used. The authors propose a new way to train these models faster and more accurately by using second-order optimization methods. They show this approach works well for detecting road surface problems based on signals from tire sensors. This method could help machines learn better from complex data to spot issues on roads.
What this means in practice
- •For vehicle safety engineers: Train better models to detect road surface abnormalities using tire sensor data with improved accuracy and efficiency.
- •For signal processing developers: Use second-order optimization to train kernel-based models more efficiently on 1D sensor signals.
Authors
Andrea Angino, Matthias Voigt, Rolf Krause, Tamás Dózsa
Abstract
We introduce a novel second-order optimization framework for minimizing so-called variable projection functionals. We demonstrate that the proposed framework is especially usefulfor the training of variable projection based kernel methods. In particular, the problem of efficiently training variable projection support vector machines (VP-SVMs) is considered. We show the effectiveness of the proposed training methodology in a real-world application, namely we demonstrate how second-order trust region algorithms can be used to train VPSVM models to recognize road surface abnormalities based on 1D signals obtained from a tire sensor.