paper-with-me

홈 › Papers

Mitigating the Likelihood Paradox in Flow-based OOD Detection via Entropy Manipulation

2026-02-10 · Donghwan Kim, Hyunsoo Yoon arxiv

Deep generative models that can tractably compute input likelihoods, including normalizing flows, often assign unexpectedly high likelihoods to out-of-distribution (OOD) inputs. We mitigate this likelihood paradox by manipulating input entropy based on semantic similarity, applying stronger perturbations to inputs that are less similar to an in-distribution memory bank. We provide a theoretical analysis showing that entropy control increases the expected log-likelihood gap between in-distribution and OOD samples in favor of the in-distribution, and we explain why the procedure works without any additional training of the density model. We then evaluate our method against likelihood-based OOD detectors on standard benchmarks and find consistent AUROC improvements over baselines, supporting our explanation.

📄 PDF Abstract BibTeX arXiv:2602.09581

Code (0)

등록된 구현이 없습니다.

Tasks

Semantic Similarity

Similar Papers 제목 키워드 기반

A Geometric Explanation of the Likelihood OOD Detection Paradox

2024-03-27 · Hamidreza Kamkari, Brendan Leigh Ross, Jesse C. Cresswell, Anthony L. Caterini 외

Likelihood-based deep generative models (DGMs) commonly exhibit a puzzling behaviour: when trained on a relatively complex dataset, they assign higher likelihood values to out-of-distribution (OOD) data from simpler sour…

Local Diagnostics of Continuous Normalizing Flow for Out-of-Distribution Detection

2026-05-30 · Xinwei Cao, Mengxuan Lu, Torbjørn Svendsen, Giampiero Salvi arxiv

We address the problem of out-of-distribution (OOD) detection for target observations embedded in a subspace of the high dimensional data space. Using continuous normalizing flows (CNFs), we propose a Lagrangian sub-flow…

Out-of-Distribution DetectionSpeech Synthesis

Joint Distillation for Fast Likelihood Evaluation and Sampling in Flow-based Models

2025-12-02 · Xinyue Ai, Yutong He, Albert Gu, Ruslan Salakhutdinov 외 arxiv

Log-likelihood evaluation enables important capabilities in generative models, including model comparison, certain fine-tuning objectives, and many downstream applications. Yet paradoxically, some of today's best generat…

Entropic Issues in Likelihood-Based OOD Detection

2021-09-22 · NeurIPS Workshop ICBINB 2021 12 · Anthony L. Caterini, Gabriel Loaiza-Ganem

Deep generative models trained by maximum likelihood remain very popular methods for reasoning about data probabilistically. However, it has been observed that they can assign higher likelihoods to out-of-distribution (O…

Out of Distribution (OOD) Detection

Cells Solved the Gibbs Paradox by Learning to Contain Entropic Forces

2023-05-17 · Josh E. Baker

As Nature's version of machine learning, evolution has solved many extraordinarily complex problems, none perhaps more remarkable than learning to harness an increase in chemical entropy (disorder) to generate directed c…