Tamara Fernandez
DPhil Statistics
About me
I am a postdoctoral research associate at the Gatsby Unit, working with Arthur Gretton. I joined the Gatsby Unit in 2018 as a Biometrika postdoctoral fellow. In 2018, I obtained my DPhil in Statistics at the University of Oxford, under the supervision of Yee Whye Teh. My work has been focused on extending modern machine learning techniques to classical statistical problems, in particular to Survival Analysis.
Contact: t.a.fernandez at ucl dot ac dot uk
Research Interests
Non-parametric statistics
Kernel methods
Survival Analysis
Bayesian methods
Publications
Under review:
A kernel test for quasi-independence. with Wenkai Xu, Marc Ditzhaus and Arthur Gretton (2020).
A kernel log-rank test of independence for right censored data. with Arthur Gretton, David Rindt and Dino Sejdinovic (2020). Submitted to: Journal of the American Statistical Association. https://arxiv.org/abs/1912.03784
A reproducing kernel Hilbert space log-rank test for the two-sample problem. with Nicolas Rivera (2019). Submitted to: Scandinavian Journal of Statistics. https://arxiv.org/pdf/1904.05187.pdf
Published:
Kernelized Stein discrepancy tests of goodness-of-fit for time-to-event data. with Wenkai Xu, Nicolas Rivera and Arthur Gretton. ICML 2020.
Kaplan Meier V- and U-statistics. with Nicolas Rivera. Electronic Journal of Statistics, 14(1):1872–1916, 2020. https://projecteuclid.org/euclid.ejs/1587693634
A maximum-mean-discrepancy goodness-of-fit test for censored data. with Arthur Gretton. AISTATS 2019. http://proceedings.mlr.press/v89/fernandez19a/fernandez19a.pdf
Gaussian Processes for Survival Analysis. with Nicolas Rivera and Yee Whye Teh. Neurips 2016. https://papers.nips.cc/paper/6443-gaussian-processes-for-survival-analysis.pdf
Technical Report:
Posterior consistency for a non-parametric survival model under a Gaussian process prior. with Yee Whye Teh 2017. https://arxiv.org/abs/1611.02335