Abstract. This paper introduces a new framework for image classification using local visual descriptors. The pipeline first performs a nonlinear feature transformation on descripto...
Abstract In case of insufficient data samples in highdimensional classification problems, sparse scatters of samples tend to have many ‘holes’—regions that have few or no nea...
Hakan Cevikalp, Diane Larlus, Marian Neamtu, Bill ...
We study optimal control problems for general unstructured nonlinear differential-algebraic equations of arbitrary index. In particular, we derive necessary conditions in the case ...
The traditional SPM approach based on bag-of-features (BoF) must use nonlinear classifiers to achieve good image classification performance. This paper presents a simple but effec...
Both example-based and model-based approaches for classifying contour shapes can encounter difficulties when dealing with classes that have large nonlinear variability, especially...