A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks.¶
Hendrycks, D., & Gimpel, K. (2017). A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks. International Conference on Learning Representations (ICLR).
Cited by¶
2 citations across 2 artifacts.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Extrapolation Beyond Sampled Regime
- Machine learning: a classifier trained on one input distribution emits high softmax confidence on out-of-distribution inputs; out-of-distribution detection exists as a field precisely because the base model's confidence apparatus is self-blind.
This sourceEstablishes out-of-distribution detection as a task precisely because a base model's softmax confidence does not register that an input lies outside the training distribution.
- Machine learning: a classifier trained on one input distribution emits high softmax confidence on out-of-distribution inputs; out-of-distribution detection exists as a field precisely because the base model's confidence apparatus is self-blind.
Mechanisms¶
- Nearest-Neighbor Assignment Rule
- Set the threshold too loose and the register stays empty not because the cover is complete but because the rule never admits a miss — the danger of out-of-distribution
This sourceShows that classifiers can assign high-confidence predictions to out-of-distribution inputs, motivating explicit OOD detection.
- Set the threshold too loose and the register stays empty not because the cover is complete but because the rule never admits a miss — the danger of out-of-distribution
Verification¶
Does it exist? Not checked yet. This entry carries no identifier to resolve. It was extracted from the citation as written in the article, normalized, and deduplicated against the rest of the registry.
Does it back the claim? Not recorded. Neither this nor any other of the 2 citations of this work carries a recorded support check.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:07e51195d891 · see in the full table