Nonparametric kernel estimators for image classification

By Barnab'as P'oczos, Liang Xiong, Danica J. Sutherland, Jeff Schneider. 2012

63 citations recorded by the source corpus.

We introduce a new discriminative learning method for image classification. We assume that the images are represented by unordered, multi-dimensional, finite sets of feature vectors, and that these sets might have different cardinality. This allows us to use consistent nonparametric divergence estimators to define new kernels over these sets, and then apply them in kernel classifiers. Our numerical results demonstrate that in many cases this approach can outperform state-of-the-art competitors on both simulated and challenging real-world datasets.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Statistical Learning Theory

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Statistical Learning Theory

Citations in the corpus

Kernel Mean Embedding of Distributions: A Review and Beyonds (2016, methodology): “In Muandet et al. (2012) and Póczos et al. (2012), SVM on distributional data was applied to natural image categorization using bag-of-word (BoW) data representation, i.e., each image is viewed as a distribution over codewords.”
Fast Distribution To Real Regression (2013, background): “Also, kernels based on nonparametric estimators of divergences have been explored [16, 14].”
On Estimating L 22 Divergence (2014, background): “We also show that this estimator is minimax optimal.”
Distribution to Distribution Regression (2013, methodology): “Futhermore, kernels based nonparametric estimators of divergences have also been explored (Póczos et al., 2012a;b).”
Beyond Covariance: Feature Representation with Nonlinear Kernel Matrices (2015, methodology): “This is different from existing work developing a kernel function to measure the similarity of two feature sets [17].”
Linear-Time Learning on Distributions with Approximate Kernel Embeddings (2015, methodology): “For example, regression from distributional covariates to real or distributional responses is possible via kernel smoothing (Póczos et al. 2012a; Oliva, Póczos, and Schneider 2013), and many learning tasks can be solved with RKHS approaches (Muandet et al. 2012; Póczos et al. 2012b).”
On Estimating $L_2^2$ Divergence (2014, background): “on distributions [16], may not require splitting the sample.”
Nonparametric Estimation of Renyi Divergence and Friends (2014, background): “It has also enjoyed success in computer vision, where features are computed for each patch of an image and these feature vectors are modeled as independent draws from an underlying distribution (Póczos et al., 2012).”