Conditional Flow Models Convert Epistemic Uncertainty into Apparent Aleatoric Variance

Key points

• Given their success in such tasks, we ask the question: what useful information do the conditional distributions produced by CFMs carry beyond what a point predictor would provide?1
• We identify a systematic phenomenon: in many natural regimes, CFM samples routinely have larger variance than the true conditional distribution, even in nearly deterministic regression problems.1
• Through controlled synthetic and real-data experiments, and comparisons with explicit smoothing estimators, we show that the induced variance tracks model underfitting of complex or hard-to-fit targets, and decreases with additional training or data.1
• We provide a theoretical account of the phenomenon in a simplified setting and demonstrate a consequent practical use-case of conditional flow-matching: designing high-coverage predictive sets that cover the ground-truth regression target.1

Sources (1)

• [1]Conditional Flow Models Convert Epistemic Uncertainty into Apparent Aleatoric Variance
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 21, 04:02 AM

  Conditional Flow Models (CFMs) are increasingly being used in place of deterministic predictors across machine learning, including for tasks that were classically formulated as regression.

  Given their success in such tasks, we ask the question: what useful information do the conditional distributions produced by CFMs carry beyond what a point predictor would provide?

Extractive summary: sentences quoted from the sources.