General Quantification of Covariate and Concept Shifts
Quick summary
arXiv:2609.11918v1 Announce Type: cross Abstract: Generalization under distribution shift remains a core challenge in modern machine learning, yet existing learning bound theory is limited to narrow, idealized settings and is non-estimable from samples. In this paper, we bridge the gap between theory and practical applications. We first show that existing definition of concept shift breaks when the source and target supports mismatch. Leveraging entropic optimal transport, we propose a key notion: $\gamma^{*}\!$-concept shifts, and derive a general error bound unifying covariate and $\gamma^{*
Key takeaways
- arXiv:2609.11918v1 Announce Type: cross Abstract: Generalization under distribution shift remains a core challenge in modern machine learning, yet existing learning bound theory is limited to narrow, idealized settings and is non-estimable from samples.
- In this paper, we bridge the gap between theory and practical applications.
- We first show that existing definition of concept shift breaks when the source and target supports mismatch.
Why it matters
The value of this work lies as much in how it was tested as in the claim itself. Sample design, baselines, uncertainty and replication help separate a laboratory result from real-world impact.
Member comments