New Benchmark Quantifies How Different Types of Data Leakage Inflate Tabular ML Results
A new arXiv paper introduces a controlled benchmark measuring how various data leakage mechanisms inflate evaluation metrics in tabular classification tasks.
Read the full story at arXiv cs.LG ↗
Timeline · 1 report
- 2026-11-17 05:00 · arXiv cs.LG
Not All Data Leakage Is Equal: A Controlled Benchmark of Evaluation Inflation in Tabular Classification