A Stable Attribution Is Not a Stable Error Rate: What Feature-Attribution Monitoring Can See Under Drift, Why 'More Sensitive Than the Inputs' and 'Silent While Fairness Fails' Are Both Correct, and the Label-Free Statistic the Fairness Alarm Needed

APA

Mahendrakar, P. (2026). A Stable Attribution Is Not a Stable Error Rate: What Feature-Attribution Monitoring Can See Under Drift, Why 'More Sensitive Than the Inputs' and 'Silent While Fairness Fails' Are Both Correct, and the Label-Free Statistic the Fairness Alarm Needed [Preprint]. Life of Research. https://doi.org/10.5281/zenodo.23871653

Copy APA

BibTeX

@misc{lor1782026,
  title        = {A Stable Attribution Is Not a Stable Error Rate: What Feature-Attribution Monitoring Can See Under Drift, Why 'More Sensitive Than the Inputs' and 'Silent While Fairness Fails' Are Both Correct, and the Label-Free Statistic the Fairness Alarm Needed},
  author       = {Pranay Mahendrakar},
  year         = {2026},
  publisher    = {Life of Research},
  doi          = {10.5281/zenodo.23871653},
  url          = {https://lifeofresearch.com/preprints/178},
  keywords     = {model monitoring, distribution shift, feature attribution, SHAP, algorithmic fairness, label-free performance estimation, concept drift, equalized odds},
  note         = {LOR-00178}
}

Copy BibTeX More on the record page