mlsanityData to 14 Oct 2026

Training-Free Spectral Transductive Refinement for Cross-Domain Few-Shot Classification

arXiv 2609.23758 ↗First posted 20 Sep 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Few-shot recognition with frozen visual features is especially fragile under domain shift and one-shot supervision, where a single labelled image is an unreliable estimate of its class. We ask how far this fragility can be reduced purely at test time, without retraining the encoder or augmenting the source domain. We present Spectral Transductive Refinement (STR), a training-free transductive inference rule that exploits the geometry of the complete support-query episode. Given frozen embeddings, STR builds a joint k-nearest-neighbour graph, maps the episode into a normalized-Laplacian spectral coordinate system, initializes class representatives from the labelled support, and iteratively refines them using pseudo-labelled queries. We evaluate STR under two protocols. A controlled component study with frozen ResNet-18 features shows that spectral refinement consistently improves over single-prototype spectral initialization across five shifted domains, with the largest gains in the one-shot regime where support estimates are weakest. We then benchmark STR against recent Cross-Domain Few-Shot Learning (CD-FSL) methods using the standard miniImageNet-pretrained ResNet-10 backbone over eight established target domains. Operating entirely at inference time, STR attains the highest 1-shot average among compared methods and remains competitive at 5-shot, rivalling approaches relying on heavy source-domain meta-training augmentations. Because STR is transductive, we report its setting explicitly. Diagnostics attribute its gains to iterative refinement in spectral coordinates rather than added prototype capacity, which remains inactive in our configuration.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 26 of its 43 references that are in the map. First posted on arXiv 20 Sep 2026.

Builds on

26 of 43 references.

References · 26

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
  3. 2017Prototypical Networks for Few-shot Learningcited by 3,015
  4. 2016Matching Networks for One Shot Learningcited by 2,255
  5. 2013Describing Textures in the Wildcited by 2,125
  6. 2017EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classificationcited by 1,496
  7. 2017ChestX-ray8: Hospital-scale Chest X-ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseasescited by 972
  8. 2018Meta-Learning for Semi-Supervised Few-Shot Classificationcited by 456
  9. 2020Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?cited by 437
  10. 2018TADAM: Task dependent adaptive metric for improved few-shot learningcited by 408
  11. 2017Few-Shot Learning with Graph Neural Networkscited by 324
  12. 2019Label Propagation for Deep Semi-supervised Learningcited by 289
14 more
  1. 2016Using Deep Learning for Image-Based Plant Disease Detectioncited by 219
  2. 2018Learning to Propagate Labels: Transductive Propagation Network for Few-shot Learningcited by 211
  3. 2020Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learningcited by 179
  4. 2022Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Differencecited by 126
  5. 2022Uncertainty Modeling for Out-of-Distribution Generalizationcited by 125
  6. 2019Infinite Mixture Prototypes for Few-Shot Learningcited by 99
  7. 2020Laplacian Regularized Few-Shot Learningcited by 92
  8. 2021Cross-Domain Few-Shot Classification via Adversarial Task Augmentationcited by 63
  9. 2023StyleAdv: Meta Style Adversarial Training for Cross-Domain Few-Shot Learningcited by 53
  10. 2020Transductive Information Maximization For Few-Shot Learningcited by 45
  11. 2023Strong Baselines for Parameter Efficient Few-Shot Fine-tuningcited by 37
  12. 2022Realistic Evaluation of Transductive Few-Shot Learningcited by 27
  13. 2024Flatten Long-Range Loss Landscapes for Cross-Domain Few-Shot Learningcited by 26
  14. 2024Meta-Exploiting Frequency Prior for Cross-Domain Few-Shot Learningcited by 11

Built on by

Not cited yet.