# MIRA: Mutual Information guided calibration for Reliable Test-Time Adaptation

DevFeed: [MIRA: Mutual Information guided calibration for Reliable Test-Time Adaptation](<https://devfeed.tech/articles/mira-mutual-information-guided-calibration-for-reliable-test-time-adaptation-114627.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.02003>)

Author: Hyeongyu Kim, Mingyeong Jang, Youngjun Song, GeonHui Han, Dosik Hwang

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Data science and engineering](<https://devfeed.tech/topics/data-science-and-engineering.md>), [Hyperparameter optimization](<https://devfeed.tech/topics/hyperparameter-optimization.md>)

Tags: [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [entropy](<https://devfeed.tech/tags/entropy.md>), [errors](<https://devfeed.tech/tags/errors.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [feature](<https://devfeed.tech/tags/feature.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [improvements](<https://devfeed.tech/tags/improvements.md>), [information](<https://devfeed.tech/tags/information.md>), [layer](<https://devfeed.tech/tags/layer.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [local](<https://devfeed.tech/tags/local.md>), [mira](<https://devfeed.tech/tags/mira.md>), [model](<https://devfeed.tech/tags/model.md>), [offline](<https://devfeed.tech/tags/offline.md>), [online](<https://devfeed.tech/tags/online.md>), [prediction](<https://devfeed.tech/tags/prediction.md>), [predictions](<https://devfeed.tech/tags/predictions.md>), [test](<https://devfeed.tech/tags/test.md>)

## AI overview

MIRA aims to reduce harmful test-time adaptation by checking how stable predictions are to small perturbations of the classifier's last layer. For each unlabeled target batch, it reuses one feature-extraction pass, samples perturbed classifiers around the source classifier, and selects a batch-specific probing scale using mutual information and prediction agreement. It then weights entropy minimization to reduce the influence of low-confidence or perturbation-sensitive predictions. The method is described as source-free and requiring no retraining, ensembles, or offline posterior fitting. The abstract reports improved accuracy, calibration, and negative log-likelihood on distribution-shift benchmarks, but provides no benchmark details or quantitative results.

## Source excerpt

arXiv:2611.02003v1 Announce Type: new Abstract: Test-time adaptation (TTA) updates a deployed model using only unlabeled target data, but standard entropy-based methods often rely on deterministic confidence and can reinforce overconfident errors under distribution shift. We propose MIRA, Mutual Information guided calibration for Reliable Test-Time Adaptation, a source-free online adaptation method that uses lightweight last-layer Gaussian perturbations to probe prediction reliability. For each target batch, MIRA reuses a single feature extraction pass and samples perturbed classifiers around the source classifier, forming an efficient local predictive distribution. Rather than fitting a Bayesian posterior, MIRA treats the perturbation scale as a local probing radius and selects a batch-specific scale using mutual information (MI) and prediction agreement. The resulting perturbation-sensitivity signal identifies predictions that are confident and locally stable under classifier perturbations, while suppressing brittle predictions with high stochastic disagreement. MIRA then performs adaptation with an MI-confidence weighted entropy loss, which down-weights perturbation-sensitive or low-confidence samples before they drive entropy minimization. This weighted loss reduces the influence of brittle predictions that can otherwise cause harmful adaptation under distribution shift. The method requires no source data, retraining, model ensembles, or offline posterior fitting, and can be applied as a plug-and-play modification to entropy-minimization TTA. Experiments on distribution-shift benchmarks show improvements in accuracy, calibration, and negative log-likelihood, suggesting that mutual-information-guided perturbation probing and weighted adaptation yield more robust online TTA than deterministic confidence.