N
Nagarajan Natarajan
Researcher at Microsoft
Publications - 41
Citations - 2594
Nagarajan Natarajan is an academic researcher from Microsoft. The author has contributed to research in topics: Binary classification & Computer science. The author has an hindex of 17, co-authored 37 publications receiving 2348 citations. Previous affiliations of Nagarajan Natarajan include University of Texas at Austin.
Papers
More filters
Proceedings Article
Learning with Noisy Labels
TL;DR: In this paper, a simple unbiased estimator of any loss is provided, and performance bounds for empirical risk minimization in the presence of iid data with noisy labels are obtained, leading to an efficient algorithm for empirical minimization.
Journal ArticleDOI
Inductive matrix completion for predicting gene-disease associations.
TL;DR: A novel matrix-completion method called Inductive Matrix Completion is applied to the problem of predicting gene-disease associations; it combines multiple types of evidence for diseases and genes to learn latent factors that explain the observed gene–diseases associations.
Proceedings ArticleDOI
Exploiting longer cycles for link prediction in signed networks
TL;DR: A supervised machine learning based link prediction method that uses features derived from longer cycles in the network that outperforms all previous approaches on 3 networks drawn from sources such as Epinions, Slashdot and Wikipedia.
Journal ArticleDOI
Prediction and Validation of Gene-Disease Associations Using Methods Inspired by Social Network Analyses
U. Martin Singh-Blom,Nagarajan Natarajan,Ambuj Tewari,John O. Woods,Inderjit S. Dhillon,Edward M. Marcotte +5 more
TL;DR: The Katz measure is better at identifying associations between traits and poorly studied genes, whereas Catapult is better suited to correctly identifying gene-trait associations overall.
Proceedings Article
Consistent Binary Classification with Generalized Performance Metrics
TL;DR: This analysis identifies the optimal classifiers as the sign of the thresholded conditional probability of the positive class, with a performance metric-dependent threshold, and proposes two algorithms for estimating them, and proves their statistical consistency.