SOCP uses self-organizing maps for unsupervised group discovery to enable local calibration in conformal prediction, reducing regional coverage gaps on benchmarks with small set-size increases while preserving validity guarantees.
Inductive Confidence Machines for Regression
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
fields
stat.ML 4years
2026 4representative citing papers
Develops a skew-adaptive split conformal prediction method that learns local skewness via a gauge-derived conformity score and an asinh residual model while preserving marginal validity under exchangeability.
Conformal Seasonal Pools is a training-free method that outperforms DeepNPTS on CRPS, quantile loss, and especially 95% coverage (0.89 vs 0.66) across six time-series datasets while being over 500x faster on CPU.
A training-free conformal naive interval based on last-value forecasts provides stronger or comparable performance to many learned methods in one-step probabilistic time series forecasting and should be required as a baseline.
citing papers explorer
-
Self-Organized Conformal Prediction: Reducing Regional Coverage Gaps with Unsupervised Group Discovery
SOCP uses self-organizing maps for unsupervised group discovery to enable local calibration in conformal prediction, reducing regional coverage gaps on benchmarks with small set-size increases while preserving validity guarantees.
-
Skew-adaptive conformal prediction
Develops a skew-adaptive split conformal prediction method that learns local skewness via a gauge-derived conformity score and an asinh residual model while preserving marginal validity under exchangeability.
-
Training-Free Probabilistic Time-Series Forecasting with Conformal Seasonal Pools
Conformal Seasonal Pools is a training-free method that outperforms DeepNPTS on CRPS, quantile loss, and especially 95% coverage (0.89 vs 0.66) across six time-series datasets while being over 500x faster on CPU.
-
Report the Floor: A Training-Free Conformal Interval Is a Mandatory Baseline for Probabilistic Time-Series Forecasting
A training-free conformal naive interval based on last-value forecasts provides stronger or comparable performance to many learned methods in one-step probabilistic time series forecasting and should be required as a baseline.