Learning to detect objects in images via a sparse, part-based representation
- S. Agarwal, A. Awan, D. Roth
- Computer ScienceIEEE Transactions on Pattern Analysis and Machineā¦
- 1 November 2004
A learning-based approach to the problem of detecting objects in still, gray-scale images that makes use of a sparse, part-based representation is developed and a critical evaluation of the approach under the proposed standards is presented.
Learning a Sparse Representation for Object Detection
- S. Agarwal, D. Roth
- Computer ScienceEuropean Conference on Computer Vision
- 28 May 2002
An approach for learning to detect objects in still gray images, that is based on a sparse, part-based representation of objects, that achieves high detection accuracy on a difficult test set of real-world images, and is highly robust to partial occlusion and background variation.
Generalization Bounds for the Area Under the ROC Curve
- S. Agarwal, T. Graepel, R. Herbrich, Sariel Har-Peled, D. Roth
- Computer Science, MathematicsJournal of machine learning research
- 1 December 2005
The expected accuracy of a ranking function is defined (analogous to the expected error rate of a classification function), and distribution-free probabilistic bounds on the deviation of the empirical AUC of aranking function (observed on a finite data sequence) are derived from its expected accuracy.
A Structural SVM Based Approach for Optimizing Partial AUC
- H. Narasimhan, S. Agarwal
- Computer ScienceInternational Conference on Machine Learning
- 16 June 2013
A structural SVM framework for directly optimizing the partial AUC between any two false positive rates and an efficient algorithm for solving this combinatorial optimization problem that has the same computational complexity as Joachims' algorithm for optimizing the usual AUC is developed.
Generalization Bounds for Ranking Algorithms via Algorithmic Stability
- S. Agarwal, P. Niyogi
- Computer ScienceJournal of machine learning research
- 1 December 2009
It is shown that kernel-based ranking algorithms that perform regularization in a reproducing kernel Hilbert space have such stability properties, and therefore bounds can be applied to these algorithms; this is in contrast with generalization bounds based on uniform convergence, which in many cases cannot be appliedTo this point, earlier results that were derived in the special setting of bipartite ranking are generalized.
The Infinite Push: A New Support Vector Ranking Algorithm that Directly Optimizes Accuracy at the Absolute Top of the List
- S. Agarwal
- Computer ScienceSDM
- 2011
A new ranking algorithm that directly maximizes the number of relevant objects retrieved at the absolute top of the list using the recent l1, ā projection method of Quattoni et al (2009).
Stability and Generalization of Bipartite Ranking Algorithms
- S. Agarwal, P. Niyogi
- Computer Science, MathematicsAnnual Conference Computational Learning Theory
- 27 June 2005
It is shown that kernel-based ranking algorithms that perform regularization in a reproducing kernel Hilbert space have such stability properties, and therefore the bounds can be applied to these algorithms; this is in contrast with previous generalization bounds for ranking, which are based on uniform convergence and in many cases cannot be appliedto these algorithms.
On the Statistical Consistency of Plug-in Classifiers for Non-decomposable Performance Measures
- H. Narasimhan, Rohit Vaish, S. Agarwal
- Computer Science, MathematicsNIPS
- 8 December 2014
This work considers plug-in algorithms that learn a classifier by applying an empirically determined threshold to a suitable 'estimate' of the class probability, and provides a general methodology to show consistency of these methods for any non-decomposable measure that can be expressed as a continuous function of true positive rate and true negative rate.
Consistent algorithms for multiclass classification with an abstain option
- H. G. Ramaswamy, Ambuj Tewari, S. Agarwal
- Computer Science
- 2018
The goal is to design consistent algorithms for such n-class classification problems with a āreject optionā; while such algorithms are known for the binary (n = 2) case, little has been understood for the general multiclass case.
Ranking on graph data
- S. Agarwal
- Computer Science, MathematicsInternational Conference on Machine Learning
- 25 June 2006
An algorithmic framework for learning ranking functions on graph data, based on recent developments in regularization theory for graphs and corresponding Laplacian-based methods for classification, is developed.
...
...