The Limitations of Model Retraining in the Face of Performativity
This addresses a foundational issue in machine learning for practitioners dealing with dynamic environments, though it is incremental as it builds on existing performativity concepts.
The paper tackles the problem of model retraining under performative shifts, where data distribution changes due to the deployed model, showing that naive retraining is suboptimal and proposing regularization to achieve optimal models.
We study stochastic optimization in the context of performative shifts, where the data distribution changes in response to the deployed model. We demonstrate that naive retraining can be provably suboptimal even for simple distribution shifts. The issue worsens when models are retrained given a finite number of samples at each retraining step. We show that adding regularization to retraining corrects both of these issues, attaining provably optimal models in the face of distribution shifts. Our work advocates rethinking how machine learning models are retrained in the presence of performative effects.