Related papers: The Mean is the Mirage: Entropy-Adaptive Model Merging under Heterogeneous Domain Shifts in Medical Imaging

The Mean is the Mirage: Entropy-Adaptive Model Merging under Heterogeneous Domain Shifts in Medical Imaging

URL: http://arxiv.org/abs/2602.21372v1
Date: Tue, 24 Feb 2026 21:06:19 GMT
Title: The Mean is the Mirage: Entropy-Adaptive Model Merging under Heterogeneous Domain Shifts in Medical Imaging
Authors: Sameer Ambekar, Reza Nasirigerdeh, Peter J. Schuffler, Lina Felsner, Daniel M. Lang, Julia A. Schnabel,
Abstract summary: Model merging under unseen test-time distribution shifts often renders naive strategies, such as mean averaging unreliable.<n>We introduce an entropy-adaptive, fully online model-merging method that yields a batch-specific merged model via only forward passes.<n>We extensively evaluate our method with state-of-the-art baselines using two backbones across nine medical and natural-domain generalization image classification datasets.
Score: 3.597779662054083
License: http://creativecommons.org/licenses/by/4.0/
Abstract: Model merging under unseen test-time distribution shifts often renders naive strategies, such as mean averaging unreliable. This challenge is especially acute in medical imaging, where models are fine-tuned locally at clinics on private data, producing domain-specific models that differ by scanner, protocol, and population. When deployed at an unseen clinical site, test cases arrive in unlabeled, non-i.i.d. batches, and the model must adapt immediately without labels. In this work, we introduce an entropy-adaptive, fully online model-merging method that yields a batch-specific merged model via only forward passes, effectively leveraging target information. We further demonstrate why mean merging is prone to failure and misaligned under heterogeneous domain shifts. Next, we mitigate encoder classifier mismatch by decoupling the encoder and classification head, merging with separate merging coefficients. We extensively evaluate our method with state-of-the-art baselines using two backbones across nine medical and natural-domain generalization image classification datasets, showing consistent gains across standard evaluation and challenging scenarios. These performance gains are achieved while retaining single-model inference at test-time, thereby demonstrating the effectiveness of our method.

Related papers

DDB: Diffusion Driven Balancing to Address Spurious Correlations [24.940576844328408]
Deep neural networks trained with Empirical Risk Minimization often fail to generalize to out-of-distribution samples.<n>We propose a Diffusion Driven Balancing (DDB) technique to generate training samples with text-to-image diffusion models.<n>Our experiments show that our technique achieves better worst-group accuracy than the existing state-of-the-art methods.
arXiv Detail & Related papers (2025-03-21T15:28:22Z)
Adaptive Face Recognition Using Adversarial Information Network [57.29464116557734]
Face recognition models often degenerate when training data are different from testing data. We propose a novel adversarial information network (AIN) to address it.
arXiv Detail & Related papers (2023-05-23T02:14:11Z)
Contrastive Model Adaptation for Cross-Condition Robustness in Semantic Segmentation [58.17907376475596]
We investigate normal-to-adverse condition model adaptation for semantic segmentation. Our method -- CMA -- leverages such image pairs to learn condition-invariant features via contrastive learning. We achieve state-of-the-art semantic segmentation performance for model adaptation on several normal-to-adverse adaptation benchmarks.
arXiv Detail & Related papers (2023-03-09T11:48:29Z)
VAESim: A probabilistic approach for self-supervised prototype discovery [0.23624125155742057]
We propose an architecture for image stratification based on a conditional variational autoencoder. We use a continuous latent space to represent the continuum of disorders and find clusters during training, which can then be used for image/patient stratification. We demonstrate that our method outperforms baselines in terms of kNN accuracy measured on a classification task against a standard VAE.
arXiv Detail & Related papers (2022-09-25T17:55:31Z)
Self-Supervised Training with Autoencoders for Visual Anomaly Detection [61.62861063776813]
We focus on a specific use case in anomaly detection where the distribution of normal samples is supported by a lower-dimensional manifold. We adapt a self-supervised learning regime that exploits discriminative information during training but focuses on the submanifold of normal examples. We achieve a new state-of-the-art result on the MVTec AD dataset -- a challenging benchmark for visual anomaly detection in the manufacturing domain.
arXiv Detail & Related papers (2022-06-23T14:16:30Z)
On-the-Fly Test-time Adaptation for Medical Image Segmentation [63.476899335138164]
Adapting the source model to target data distribution at test-time is an efficient solution for the data-shift problem. We propose a new framework called Adaptive UNet where each convolutional block is equipped with an adaptive batch normalization layer. During test-time, the model takes in just the new test image and generates a domain code to adapt the features of source model according to the test data.
arXiv Detail & Related papers (2022-03-10T18:51:29Z)
on the effectiveness of generative adversarial network on anomaly detection [1.6244541005112747]
GANs rely on the rich contextual information of these models to identify the actual training distribution. We suggest a new unsupervised model based on GANs --a combination of an autoencoder and a GAN. A new scoring function was introduced to target anomalies where a linear combination of the internal representation of the discriminator and the generator's visual representation, plus the encoded representation of the autoencoder, come together to define the proposed anomaly score.
arXiv Detail & Related papers (2021-12-31T16:35:47Z)
Training on Test Data with Bayesian Adaptation for Covariate Shift [96.3250517412545]
Deep neural networks often make inaccurate predictions with unreliable uncertainty estimates. We derive a Bayesian model that provides for a well-defined relationship between unlabeled inputs under distributional shift and model parameters. We show that our method improves both accuracy and uncertainty estimation.
arXiv Detail & Related papers (2021-09-27T01:09:08Z)
Consistent Posterior Distributions under Vessel-Mixing: A Regularization for Cross-Domain Retinal Artery/Vein Classification [30.30848090813239]
We propose a vessel-mixing based consistency regularization framework, for cross-domain learning in retinal A/V classification. Our method achieves the state-of-the-art cross-domain performance, which is also close to the upper bound obtained by fully supervised learning on target domain.
arXiv Detail & Related papers (2021-03-16T14:18:35Z)
Semi-supervised Medical Image Classification with Relation-driven Self-ensembling Model [71.80319052891817]
We present a relation-driven semi-supervised framework for medical image classification. It exploits the unlabeled data by encouraging the prediction consistency of given input under perturbations. Our method outperforms many state-of-the-art semi-supervised learning methods on both single-label and multi-label image classification scenarios.
arXiv Detail & Related papers (2020-05-15T06:57:54Z)

This list is automatically generated from the titles and abstracts of the papers in this site.