Data Selection for ERMs
This work addresses data efficiency for machine learning practitioners by offering a data-centric approach to reduce training costs while maintaining performance, though it is incremental as it builds on existing learning theory.
The paper tackles the problem of selecting a small subset of training data for empirical risk minimizers to achieve performance comparable to using the entire dataset, providing optimal bounds for tasks like mean estimation, linear classification, and linear regression.
Learning theory has traditionally followed a model-centric approach, focusing on designing optimal algorithms for a fixed natural learning task (e.g., linear classification or regression). In this paper, we adopt a complementary data-centric perspective, whereby we fix a natural learning rule and focus on optimizing the training data. Specifically, we study the following question: given a learning rule $\mathcal{A}$ and a data selection budget $n$, how well can $\mathcal{A}$ perform when trained on at most $n$ data points selected from a population of $N$ points? We investigate when it is possible to select $n \ll N$ points and achieve performance comparable to training on the entire population. We address this question across a variety of empirical risk minimizers. Our results include optimal data-selection bounds for mean estimation, linear classification, and linear regression. Additionally, we establish two general results: a taxonomy of error rates in binary classification and in stochastic convex optimization. Finally, we propose several open questions and directions for future research.