mlsanityData to 7 Oct 2026

Diagnosing and Remedying Shot Sensitivity with Cosine Few-Shot Learners

arXiv 2207.03398 ↗First posted 7 Jul 2022cs.CV cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Few-shot recognition involves training an image classifier to distinguish novel concepts at test time using few examples (shot). Existing approaches generally assume that the shot number at test time is known in advance. This is not realistic, and the performance of a popular and foundational method has been shown to suffer when train and test shots do not match. We conduct a systematic empirical study of this phenomenon. In line with prior work, we find that shot sensitivity is broadly present across metric-based few-shot learners, but in contrast to prior work, larger neural architectures provide a degree of built-in robustness to varying test shot. More importantly, a simple, previously known but greatly overlooked class of approaches based on cosine distance consistently and greatly improves robustness to shot variation, by removing sensitivity to sample noise. We derive cosine alternatives to popular and recent few-shot classifiers, broadening their applicability to realistic settings. These cosine models consistently improve shot-robustness, outperform prior shot-robust state of the art, and provide competitive accuracy on a range of benchmarks and architectures, including notable gains in the very-low-shot regime.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 27 of its 37 references that are in the map. First posted on arXiv 7 Jul 2022.

Builds on

27 of 37 references.

References · 27

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  3. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  4. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,617
  5. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,861
  6. 2017Prototypical Networks for Few-shot Learningcited by 3,001
  7. 2016Matching Networks for One Shot Learningcited by 2,246
  8. 2019LVIS: A Dataset for Large Vocabulary Instance Segmentationcited by 1,215
  9. 2018On First-Order Meta-Learning Algorithmscited by 1,003
  10. 2019A Closer Look at Few-shot Classificationcited by 679
  11. 2018Meta-Learning with Latent Embedding Optimizationcited by 476
  12. 2019Generalizing from a Few Examples: A Survey on Few-Shot Learningcited by 462
15 more
  1. 2019Meta-Learning with Differentiable Convex Optimizationcited by 461
  2. 2018Meta-Learning for Semi-Supervised Few-Shot Classificationcited by 456
  3. 2020Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?cited by 435
  4. 2018Meta-learning with differentiable closed-form solverscited by 395
  5. 2018Dynamic Few-Shot Visual Learning without Forgettingcited by 385
  6. 2018Few-Shot Learning via Embedding Adaptation with Set-to-Set Functionscited by 293
  7. 2019Hyperbolic Image Embeddingscited by 219
  8. 2019SimpleShot: Revisiting Nearest-Neighbor Classification for Few-Shot Learningcited by 194
  9. 2020Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learningcited by 179
  10. 2020Few-Shot Classification with Feature Map Reconstruction Networkscited by 116
  11. 2020Unraveling Meta-Learning: Understanding Feature Representations for Few-Shot Taskscited by 47
  12. 2020On Episodes, Prototypical Networks, and Few-shot Learningcited by 46
  13. 2019A Theoretical Analysis of the Number of Shots in Few-Shot Learningcited by 39
  14. 2019Few-Shot Learning with Localization in Realistic Settingscited by 34
  15. 2020Revisiting Pose-Normalization for Fine-Grained Few-Shot Recognitioncited by 30

Built on by

Not cited yet.