Confidence Calibration under Ambiguous Ground Truth
About
Confidence calibration assumes a unique ground-truth label per input, yet this assumption fails wherever annotators genuinely disagree. Post-hoc calibrators fitted on majority-voted labels, the standard single-label targets used in practice, can appear well-calibrated under conventional evaluation yet remain substantially miscalibrated against the underlying annotator distribution. We show that this failure is structural: under simplifying assumptions, Temperature Scaling is biased toward temperatures that underestimate annotator uncertainty, with true-label miscalibration increasing monotonically with annotation entropy. To address this, we develop a family of ambiguity-aware post-hoc calibrators that optimise proper scoring rules against the full label distribution and require no model retraining. Our methods span progressively weaker annotation requirements: Dirichlet-Soft leverages the full annotator distribution and achieves the best overall calibration quality across settings; Monte Carlo Temperature Scaling with a single annotation per example (MCTS S=1) matches full-distribution calibration across all benchmarks, demonstrating that pre-aggregated label distributions are unnecessary; and Label-Smooth Temperature Scaling (LS-TS) operates with voted labels alone by constructing data-driven pseudo-soft targets from the model's own confidence. Experiments on four benchmarks with real multi-annotator distributions (CIFAR-10H, ChaosNLI) and clinically-informed synthetic annotations (ISIC~2019, DermaMNIST) show that Dirichlet-Soft reduces true-label ECE by 55-87% relative to Temperature Scaling, while LS-TS reduces ECE by 9-77% without any annotator data.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Medical Image Classification | ISIC 2019 m=9 synthetic annotators, K=8, mean agreement 75% (test) | ECE (Expected Calibration Error)1.77 | 26 | |
| Medical Image Classification | DermaMNIST m=5 synthetic annotators, K=7, mean agreement 64.7% (test) | ECE2.06 | 26 | |
| Probability Calibration | ChaosNLI combined SNLI+MNLI K=3 | ECE2.15 | 26 | |
| Probability Calibration | CIFAR-10H K=10 (test) | ECE0.72 | 26 |