no code implementations • 12 Oct 2023 • Numair Sani, Atalanti A. Mastakouri
Existing work to further tighten these bounds by leveraging extra information either provides numerical bounds, symbolic bounds for fixed dimensionality, or requires access to multiple datasets that contain the same treatment and outcome variables.
no code implementations • 4 Apr 2023 • Numair Sani, Atalanti A. Mastakouri, Dominik Janzing
In the absence of such assumptions, existing work requires multiple observations of datasets that contain the same treatment and outcome variables, in order to establish bounds on these probabilities.
no code implementations • 1 Dec 2021 • Nathan Drenkow, Numair Sani, Ilya Shpitser, Mathias Unberath
We find this area of research has received disproportionately less attention relative to adversarial machine learning, yet a significant robustness gap exists that manifests in performance degradation similar in magnitude to adversarial conditions.
no code implementations • 19 May 2021 • Numair Sani, Yizhen Xu, AmirEmad Ghassami, Ilya Shpitser
For binary treatments, efficient estimators for the direct and indirect effects are presented in Tchetgen Tchetgen and Shpitser (2012) based on the influence function of the parameter of interest.
no code implementations • 8 Jun 2020 • Numair Sani, Jaron Lee, Razieh Nabi, Ilya Shpitser
In order to combat this shortcoming, we propose a novel approach to trading off interpretability and performance in prediction models using ideas from semiparametric statistics, allowing us to combine the interpretability of parametric regression models with performance of nonparametric methods.
no code implementations • 3 Jun 2020 • Numair Sani, Daniel Malinsky, Ilya Shpitser
However, existing approaches have two important shortcomings: (i) the "explanatory units" are micro-level inputs into the relevant prediction model, e. g., image pixels, rather than interpretable macro-level features that are more useful for understanding how to possibly change the algorithm's behavior, and (ii) existing approaches assume there exists no unmeasured confounding between features and target model predictions, which fails to hold when the explanatory units are macro-level variables.