Kevin Maik Jablonka

New metric 'prediction churn' highlights ML model instability

Researchers have identified a new metric called "cross-sample prediction churn" to measure the instability of machine learning models in scientific applications. This metric quantifies how predictions change when differ…