Hopfield Boosting for Out-of-Distribution Detection

Claus Hofmann; Simon Lucas Schmid; Bernhard Lehner; Daniel Klotz; Sepp Hochreiter

Hopfield Boosting for Out-of-Distribution Detection

Claus Hofmann, Simon Lucas Schmid, Bernhard Lehner, Daniel Klotz, Sepp Hochreiter

Embedded AI (eAI)

Research output: Conference proceeding/Chapter in Book/Report/ › Conference Paper › peer-review

Abstract

Out-of-distribution (OOD) detection is crucial for real-world machine learning. Outlier exposure methods, which use auxiliary outlier data, can significantly enhance OOD detection. We present Hopfield Boosting, a boosting technique employing modern Hopfield energy (MHE) to refine the boundary between in-distribution (ID) and OOD data. Our method focuses on challenging outlier examples near the decision boundary, achieving a 40% improvement in FPR95 on CIFAR-10, setting a new OOD detection state-of-the-art with outlier exposure.

Original language	English
Title of host publication	NeurIPS 2023 Workshop
Subtitle of host publication	Associative Memory & Hopfield Networks
Publication status	Accepted/In press - 15 Dec 2023

Keywords

hopfield networks, hopfield energy, ood, out of distribution, boosting, maximum margin, deep learning

Access to Document

42_hopfield_boosting_for_out_of_d

Cite this

@inproceedings{53cc6c5387af4def90cb9dd0e10c5b7f,

title = "Hopfield Boosting for Out-of-Distribution Detection",

abstract = "Out-of-distribution (OOD) detection is crucial for real-world machine learning. Outlier exposure methods, which use auxiliary outlier data, can significantly enhance OOD detection. We present Hopfield Boosting, a boosting technique employing modern Hopfield energy (MHE) to refine the boundary between in-distribution (ID) and OOD data. Our method focuses on challenging outlier examples near the decision boundary, achieving a 40% improvement in FPR95 on CIFAR-10, setting a new OOD detection state-of-the-art with outlier exposure.",

keywords = "hopfield networks, hopfield energy, ood, out of distribution, boosting, maximum margin, deep learning",

author = "Claus Hofmann and Schmid, {Simon Lucas} and Bernhard Lehner and Daniel Klotz and Sepp Hochreiter",

year = "2023",

month = dec,

day = "15",

language = "English",

booktitle = "NeurIPS 2023 Workshop",

}

TY - GEN

T1 - Hopfield Boosting for Out-of-Distribution Detection

AU - Hofmann, Claus

AU - Schmid, Simon Lucas

AU - Lehner, Bernhard

AU - Klotz, Daniel

AU - Hochreiter, Sepp

PY - 2023/12/15

Y1 - 2023/12/15

N2 - Out-of-distribution (OOD) detection is crucial for real-world machine learning. Outlier exposure methods, which use auxiliary outlier data, can significantly enhance OOD detection. We present Hopfield Boosting, a boosting technique employing modern Hopfield energy (MHE) to refine the boundary between in-distribution (ID) and OOD data. Our method focuses on challenging outlier examples near the decision boundary, achieving a 40% improvement in FPR95 on CIFAR-10, setting a new OOD detection state-of-the-art with outlier exposure.

AB - Out-of-distribution (OOD) detection is crucial for real-world machine learning. Outlier exposure methods, which use auxiliary outlier data, can significantly enhance OOD detection. We present Hopfield Boosting, a boosting technique employing modern Hopfield energy (MHE) to refine the boundary between in-distribution (ID) and OOD data. Our method focuses on challenging outlier examples near the decision boundary, achieving a 40% improvement in FPR95 on CIFAR-10, setting a new OOD detection state-of-the-art with outlier exposure.

KW - hopfield networks, hopfield energy, ood, out of distribution, boosting, maximum margin, deep learning

M3 - Conference Paper

BT - NeurIPS 2023 Workshop

ER -

Hopfield Boosting for Out-of-Distribution Detection

Abstract

Keywords

Access to Document

Fingerprint

Cite this