How to Learn when Data Reacts to Your Model: Performative Gradient Descent
This addresses a critical issue in machine learning for applications like finance, where models interact with data, and is a foundational advance beyond incremental improvements.
The paper tackles the problem of performative distribution shift, where model deployment changes the data distribution, by introducing performative gradient descent (PerfGD), which provably converges to the optimal point, unlike prior methods that only find stable points.
Performative distribution shift captures the setting where the choice of which ML model is deployed changes the data distribution. For example, a bank which uses the number of open credit lines to determine a customer's risk of default on a loan may induce customers to open more credit lines in order to improve their chances of being approved. Because of the interactions between the model and data distribution, finding the optimal model parameters is challenging. Works in this area have focused on finding stable points, which can be far from optimal. Here we introduce performative gradient descent (PerfGD), which is the first algorithm which provably converges to the performatively optimal point. PerfGD explicitly captures how changes in the model affects the data distribution and is simple to use. We support our findings with theory and experiments.