Mini-batch bagging and attribute ranking for accurate user authentication in keystroke dynamics
We have proposed mini-batch bagging (MINIBAG) method and attribute ranking of one-class nave Bayes (AR-ONENB) algorithm.We have presented attribute-by-attribute data fragmentation technique which is used in MINIBAG method.MINIBAG facilitates machine ...
The connected-component labeling problem
Connected-component labeling (CCL) is indispensable for pattern recognition.Many connected-component labeling algorithms have been proposed.The state-of-the-art CCL algorithms presented in the last decade are reviewed. This article addresses the ...
Class Switching according to Nearest Enemy Distance for learning from highly imbalanced data-sets
We designed and implemented a novel ensemble based on Class-Switching to deal with the imbalanced class problem.The ensemble SwitchingNED changes a fraction of instances of the majority class to the minority class following a new section method based on ...
Hierarchical Multi-label Classification using Fully Associative Ensemble Learning
Developing a local hierarchical ensemble framework for Hierarchical Multi-label Classification (HMC), in which all the structural relationships in the class hierarchy are used to obtain global prediction.Introducing empirical loss minimization into HMC, ...
TextProposals
We present a text specific object proposals algorithm.Our algorithm is able to reach impressive recall rates with a few thousand proposals in different standard datasets.Our method generates word proposals without an explicit character segmentation.The ...
Learning features for offline handwritten signature verification using deep convolutional neural networks
We propose formulations for learning features for Offline Signature Verification.A novel method that uses knowledge of forgeries from a subset of users is proposed.Learned features are used to train classifiers for other users (without forgeries)...
An efficient temporal distortion measure of videos based on spacetime texture
It introduces spacetime texture for VQA as a good alternative to optical flow.It unifies motion tuning and visual saliency on spacetime texture for VQA.It has high correlation with subjective quality.It is consistent across various distortion types and ...
Feature selection for regression problems based on the Morisita estimator of intrinsic dimension
A new supervised filter for regression problems is proposed.The filter uses the newly introduced Morisita estimator of intrinsic dimension.The filter distinguishes between relevant, irrelevant and redundant features.The filter is comprehensively ...
Multiple kernel learning with hybrid kernel alignment maximization
The local kernel alignment between two kernels is first proposed.Hybrid kernel alignment combining the global and local information is designed.An alternative algorithm based on hybrid kernel alignment is proposed. Two-stage multiple kernel learning (...
Simpler editing of graph-based segmentation hierarchies using zipping algorithms
We present reusable algorithms to simplify segmentation hierarchy editing.They allow hierarchical segmentation techniques to make better use of user input.We show the use of our algorithms for non-sibling node merging and parent switching.Non-sibling ...
Graph regularized nonnegative sparse coding using incoherent dictionary for approximate nearest neighbor search
Graph Laplacian regularization is considered for improvement.Non-negativity constraints are adopted for enriching sparse codes.An iterative algorithm is proposed to learn incoherent dictionary.Interesting improvement is reaped without sacrificing query ...