Conformal e-prediction for change detection

06/03/2020

∙

We adapt conformal e-prediction to change detection, defining analogues of the Shiryaev-Roberts and CUSUM procedures for detecting violations of the IID assumption. Asymptotically, the frequency of false alarms for these analogues does not exceed the usual bounds.

READ FULL TEXT

Conformal e-prediction for change detection

Change Detection of Markov Kernels with Unknown Post Change Kernel using Maximum Mean Discrepancy

Misspecified and Asymptotically Minimax Robust Quickest Change Diagnosis

Asymptotically Optimal Pointwise and Minimax Changepoint Detection for General Stochastic Models With a Composite Post-Change Hypothesis

Quickest Change Detection with Leave-one-out Density Estimation

Sequential change detection via backward confidence sequences

E-detectors: a nonparametric framework for online changepoint detection

Power and limitations of conformal martingales

Conformal e-prediction for change detection

Related Research

Change Detection of Markov Kernels with Unknown Post Change Kernel using Maximum Mean Discrepancy

Misspecified and Asymptotically Minimax Robust Quickest Change Diagnosis

Asymptotically Optimal Pointwise and Minimax Changepoint Detection for General Stochastic Models With a Composite Post-Change Hypothesis

Quickest Change Detection with Leave-one-out Density Estimation

Sequential change detection via backward confidence sequences

E-detectors: a nonparametric framework for online changepoint detection

Power and limitations of conformal martingales