(my philosophy: academic honesty for websites includes avoiding being misleading to an outsider; this involves declaring which papers were double or single blind peer-reviewed at conferences/journals with low acceptance rates, as opposed to being accepted for submission at places with very high acceptance rates; it also includes declaring on which work you are a primary contributor of content, where you know and can discuss the paper details confidently if contacted)

* indicates a paper on which I was one of the primary contributors, ** indicates an equally contributing student author


  • Simultaneously uncovering brain regions involved in story reading subprocesses (PLoS ONE '14)
    Public Library of Science ONE, November Issue, 2014
    Leila Wehbe, B. Murphy, P. Talukdar, A. Fyshe, Aaditya Ramdas, T. Mitchell [website] [PLOS] [pdf] [supp]

  • *Fast & Flexible ADMM Algorithms for Trend Filtering
    Journal of Computational and Graphical Statistics (JCGS) - in submission
    Aaditya Ramdas, Ryan Tibshirani [arxiv] [R package] [50 min. talk]

  • *Regularized Brain Reading with Shrinkage and Smoothing
    Annals of Applied Statistics (AoAS) - in submission
    Leila Wehbe, Aaditya Ramdas, Rebecca Steorts, Cosma Shalizi [arxiv]

  • *Towards A Deeper Geometric, Analytic and Algorithmic Understanding of Margins
    Optimization Methods and Software (OMS) - in submission
    Aaditya Ramdas, Javier Pena [arxiv]


  • *Free Lunches and Computation-Statistics Tradeoffs for High Dimensional Two Sample Testing
    (in preparation)
    Aaditya Ramdas, Sashank Reddi**, Barnabas Poczos, Aarti Singh, Larry Wasserman

  • *Stein Shrinkage for Cross-Covariance Operators and Kernel Independence Testing
    (in preparation)
    Aaditya Ramdas, **Leila Wehbe [arxiv]

  • *Rows vs Columns for Randomized Ridge Regression: Kaczmarz vs Coordinate Descent
    (in preparation)
    Aaditya Ramdas, Ahmed Hefny [arxiv]

  • *Randomized Extended Gauss-Seidel : convergence in the undercomlete setting
    (in preparation)
    Aaditya Ramdas, Deanna Needell, Anna Ma

  • Fast Two-Sample Tests with Random Features
    (in preparation)
    Kacper Chwialkowski, Aaditya Ramdas, Dino Sejdinovic, Arthur Gretton


full papers, respected venues, low acceptance rates, published proceedings, single/double blind detailed reviews

  • *On the Power of a Linear-Time Nonparametric Two Sample Test in High Dimensions (AISTATS '15)
    18th International Conference on Artificial Intelligence and Statistics, San Diego, 2014
    Aaditya Ramdas, **Sashank Reddi, Aarti Singh, Barnabas Poczos, Larry Wasserman [arxiv]

  • *On the Decreasing Power of Kernel- and Distance-based Hypothesis Tests in High Dimensions (AAAI '15)
    29th AAAI Conference on Artifical Intelligence, Austin, 2015
    Aaditya Ramdas, **Sashank Reddi, Barnabas Poczos, Aarti Singh, Larry Wasserman [arxiv]

  • *Margins, Kernels and Non-linear Smoothed Perceptrons (ICML '14)
    31st International Conference on Machine Learning, Beijing, 2014
    Aaditya Ramdas, Javier Pena [ICML] [pdf] [supp] [20-min oral] [abs]

  • *An Analysis of Active Learning with Uniform Feature Noise (AISTATS '14)
    17th International Conference on Artificial Intelligence and Statistics, Reykjavik, 2014
    Aaditya Ramdas, Aarti Singh, Larry Wasserman, Barnabas Poczos [AISTATS] [pdf] [supp] [25-min oral] [abs]

  • *Algorithmic Connections Between Active Learning and Stochastic Convex Optimization (ALT '13)
    24th International Conference on Algorithmic Learning Theory, Singapore, 2013
    Aaditya Ramdas, Aarti Singh [ALT] [pdf] [25-min oral] [abs]

  • *Optimal Rates for Stochastic Convex Optimization using Tsybakov's Noise Condition (ICML '13)
    30th International Conference on Machine Learning, Atlanta, 2013
    Aaditya Ramdas, Aarti Singh [arxiv] [ICML] [pdf] [supp] [20-min oral] [abs]