More Like this

1. Robust kernels for robust location estimation

   https://doi.org/10.1016/j.neucom.2020.10.090  

   Gallego, Joseph A.; González, Fabio A.; Nasraoui, Olfa (March 2021, Neurocomputing)

   null (Ed.)
2. Robust Grouped Variable Selection Using Distributionally Robust Optimization

   https://doi.org/10.1007/s10957-022-02065-4  

   Chen, Ruidi; Paschalidis, Ioannis Ch. (September 2022, Journal of Optimization Theory and Applications)
3. Robust XVA

   https://doi.org/10.1111/mafi.12260  

   Bichuch, Maxim; Capponi, Agostino; Sturm, Stephan (March 2020, Mathematical Finance)

   Abstract We introduce an arbitrage‐free framework for robust valuation adjustments. An investor trades a credit default swap portfolio with a risky counterparty, and hedges credit risk by taking a position in defaultable bonds. The investor does not know the exact return rate of her counterparty's bond, but she knows it lies within an uncertainty interval. We derive both upper and lower bounds for the XVA process of the portfolio, and show that these bounds may be recovered as solutions of nonlinear ordinary differential equations. The presence of collateralization and closeout payoffs leads to important differences with respect to classical credit risk valuation. The value of the super‐replicating portfolio cannot be directly obtained by plugging one of the extremes of the uncertainty interval in the valuation equation, but rather depends on the relation between the XVA replicating portfolio and the closeout value throughout the life of the transaction. Our comparative statics analysis indicates that credit contagion has a nonlinear effect on the replication strategies and on the XVA. 

   more » « less
4. Robust AC Optimal Power Flow with Robust Convex Restriction

   https://doi.org/10.1109/TPWRS.2021.3075925  

   Lee, Dongchan; Turitsyn, Konstantin; Molzahn, Daniel Kenneth; Roald, Line (January 2021, IEEE Transactions on Power Systems)
5. Doubly Robust Distributionally Robust Off-Policy Evaluation and Learning

   Nathan Kallus, Xiaojie Mao (January 2022, International Conference on Machine Learning)

   Off-policy evaluation and learning (OPE/L) use offline observational data to make better decisions, which is crucial in applications where online experimentation is limited. However, depending entirely on logged data, OPE/L is sensitive to environment distribution shifts — discrepancies between the data-generating environment and that where policies are deployed. Si et al., (2020) proposed distributionally robust OPE/L (DROPE/L) to address this, but the proposal relies on inverse-propensity weighting, whose estimation error and regret will deteriorate if propensities are nonparametrically estimated and whose variance is suboptimal even if not. For standard, non-robust, OPE/L, this is solved by doubly robust (DR) methods, but they do not naturally extend to the more complex DROPE/L, which involves a worst-case expectation. In this paper, we propose the first DR algorithms for DROPE/L with KL-divergence uncertainty sets. For evaluation, we propose Localized Doubly Robust DROPE (LDR2OPE) and show that it achieves semiparametric efficiency under weak product rates conditions. Thanks to a localization technique, LDR2 OPE only requires fitting a small number of regressions, just like DR methods for standard OPE. For learning, we propose Continuum Doubly Robust DROPL (CDR2OPL) and show that, under a product rate condition involving a continuum of regressions, it enjoys a fast regret rate of 𝑂(𝑁−1/2) even when unknown propensities are nonparametrically estimated. We empirically validate our algorithms in simulations and further extend our results to general 𝑓-divergence uncertainty sets. 

   more » « less