Abstract:The contamination detection problem aims to determine whether a set of observations has been contaminated, i.e. whether it contains points drawn from a distribution different from the reference distribution. Here, we consider a supervised problem, where labeled samples drawn from both the reference distribution and the contamination distribution are available at training time. This problem is motivated by the detection of rare cells in flow cytometry. Compared to novelty detection problems or two-sample testing, where only samples from the reference distribution are available, the challenge lies in efficiently leveraging the observations from the contamination detection to design more powerful tests. In this article, we introduce a test for the supervised contamination detection problem. We provide non-asymptotic guarantees on its Type I error, and characterize its detection rate. The test relies on estimating reference and contamination densities using histograms, and its power depends strongly on the choice of the corresponding partition. We present an algorithm for judiciously choosing the partition that results in a powerful test. Simulations illustrate the good empirical performances of our partition selection algorithm and the efficiency of our test. Finally, we showcase our method and apply it to a real flow cytometry dataset.

Nonparametric Estimation for Contamination Distribution

Nonparametric Estimation of Second-Order Jump-Diffusion Model

Entropy Estimation of Contaminated Normal Distribution

Hypothesis Testing with Estimation-Dependent Costs and ∊-Contamination Prior

Asymptotic Theory for Explosive Fractional Ornstein-Uhlenbeck Processes

Choice of Estimation of Unknown Parameter under Contaminated Error Model

CONTAMINATION RATE ESTIMATION BASED ON ENTROPY

Robust Estimation under Heavy Contamination using Enlarged Models

Tractable and Near-Optimal Adversarial Algorithms for Robust Estimation in Contaminated Gaussian Models

Models of Contamination Distribution in Food Based on EM Algorithm

Adaptive Robust Confidence Intervals

Reliability Analysis Based on Progressively First-Failure-Censored Samples for the Proportional Hazard Rate Model

Consistency of Change Point Estimators for Symmetrical Stable Distribution with Parameters Shift

Estimation for Two Parameter Exponential Populations with Partially Missing Data

Asymptotic Properties of Nonparametric M-Estimation for Mixing Functional Data

A Sparse Grid Based Bayesian Method for Contaminant Source Identification

Bayes Analysis of Two-Parameter Pareto Distribution under Progressively First-Failure-censored Data

Privacy Guarantees in Posterior Sampling under Contamination

Supervised Contamination Detection, with Flow Cytometry Application

Proportional closeness estimation of probability of contamination under group testing

Posterior Robustness with Milder Conditions: Contamination Models Revisited