Distributionally Robust Performative Prediction
This work addresses robustness in performative prediction for ML systems where model deployment affects data, offering a solution to distribution map misspecification, though it is incremental as it builds on existing performative prediction concepts.
The paper tackles the problem of performative prediction where model predictions influence the data distribution, and introduces a distributionally robust framework (DRPO) to handle misspecification in the distribution map, showing provable guarantees and experimental advantages over traditional methods.
Performative prediction aims to model scenarios where predictive outcomes subsequently influence the very systems they target. The pursuit of a performative optimum (PO) -- minimizing performative risk -- is generally reliant on modeling of the distribution map, which characterizes how a deployed ML model alters the data distribution. Unfortunately, inevitable misspecification of the distribution map can lead to a poor approximation of the true PO. To address this issue, we introduce a novel framework of distributionally robust performative prediction and study a new solution concept termed as distributionally robust performative optimum (DRPO). We show provable guarantees for DRPO as a robust approximation to the true PO when the nominal distribution map is different from the actual one. Moreover, distributionally robust performative prediction can be reformulated as an augmented performative prediction problem, enabling efficient optimization. The experimental results demonstrate that DRPO offers potential advantages over traditional PO approach when the distribution map is misspecified at either micro- or macro-level.