AI & Functional Safety: Uncertainty Quantification for Safety Decisions
- Date
- 2026-08-24
- Location
- Online
- Host
- Zach L
About this event
A live 30-minute expert session on Uncertainty Quantification for Safety Decisions (AI / ML — AI & Machine Learning Safety). What We'll Cover: Aleatoric vs epistemic uncertainty and where uncertainty estimation sits in the perception-to-decision chain Methods step by step: MC dropout, deep ensembles, conformal prediction — with the calibration checks each needs How calibrated uncertainty maps to acceptance criteria and dynamic risk thresholds, producing calibration reports as evidence Common mistakes: treating raw softmax scores as probabilities and skipping recalibration after retraining The reliability diagrams, ECE metrics and threshold rationale an auditor reviews before crediting uncertainty-based safety logic Related topics: uncertainty quantification · epistemic uncertainty · aleato