mlsanityData to 14 Oct 2026

Logits DeConfusion with CLIP for Few-Shot Learning

arXiv 2504.12104 ↗First posted 16 Apr 2025cs.CVcited by 11

Not on the map. Why?

Abstract (arXiv)

With its powerful visual-language alignment capability, CLIP performs well in zero-shot and few-shot learning tasks. However, we found in experiments that CLIP's logits suffer from serious inter-class confusion problems in downstream tasks, and the ambiguity between categories seriously affects the accuracy. To address this challenge, we propose a novel method called Logits DeConfusion, which effectively learns and eliminates inter-class confusion in logits by combining our Multi-level Adapter Fusion (MAF) module with our Inter-Class Deconfusion (ICD) module. Our MAF extracts features from different levels and fuses them uniformly to enhance feature representation. Our ICD learnably eliminates inter-class confusion in logits with a residual structure. Experimental results show that our method can significantly improve the classification performance and alleviate the inter-class confusion problem. The code is available at https://github.com/LiShuo1001/LDC.

Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Builds on 48 of its 78 references that are in the map. First posted on arXiv 16 Apr 2025.

Builds on

48 of 78 references.

References · 48

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  4. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  5. 2016Feature Pyramid Networks for Object Detectioncited by 5,036
  6. 2021Learning to Prompt for Vision-Language Modelscited by 2,281
  7. 2013Describing Textures in the Wildcited by 2,125
  8. 2013Fine-Grained Visual Classification of Aircraftcited by 1,726
  9. 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,515
  10. 2017EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classificationcited by 1,496
  11. 2022Conditional Prompt Learning for Vision-Language Modelscited by 1,433
  12. 2021CLIP-Adapter: Better Vision-Language Models with Feature Adapterscited by 1,026
36 more
  1. 2019Learning Robust Global Representations by Penalizing Local Predictive Powercited by 1,013
  2. 2023Vision-Language Models for Vision Tasks: A Surveycited by 839
  3. 2022MaPLe: Multi-modal Prompt Learningcited by 687
  4. 2021LiT: Zero-Shot Transfer with Locked-image text Tuningcited by 577
  5. 2023Florence-2: Advancing a Unified Representation for a Variety of Vision Taskscited by 410
  6. 2022Tip-Adapter: Training-free Adaption of CLIP for Few-shot Classificationcited by 362
  7. 2018Meta-Transfer Learning for Few-Shot Learningcited by 306
  8. 2017Zero-Shot Learning -- The Good, the Bad and the Uglycited by 255
  9. 2022ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentationcited by 178
  10. 2022A Comprehensive Survey of Few-shot Learning: Evolution, Applications, Challenges, and Opportunitiescited by 163
  11. 2020A Review of Generalized Zero-Shot Learning Methodscited by 159
  12. 2016Learning a Deep Embedding Model for Zero-Shot Learningcited by 153
  13. 2022SuS-X: Training-Free Name-Only Transfer of Vision-Language Modelscited by 119
  14. 2017Generalized Zero-Shot Learning via Synthesized Examplescited by 114
  15. 2022CLIP Models are Few-shot Learners: Empirical Studies on VQA and Visual Entailmentcited by 113
  16. 2023LoCoOp: Few-Shot Out-of-Distribution Detection via Prompt Learningcited by 107
  17. 2019A Survey of Deep Learning-based Object Detectioncited by 106
  18. 2022CALIP: Zero-Shot Enhancement of CLIP with Parameter-free Attentioncited by 103
  19. 2023Not All Features Matter: Enhancing Few-shot CLIP with Adaptive Prior Refinementcited by 84
  20. 2024Understanding and Mitigating Language Confusion in LLMscited by 75
  21. 2024LP++: A Surprisingly Strong Linear Probe for Few-Shot CLIPcited by 60
  22. 2023Read-only Prompt Optimization for Vision-Language Few-shot Learningcited by 58
  23. 2022UniCLIP: Unified Framework for Contrastive Language-Image Pre-trainingcited by 54
  24. 2023Improving Zero-Shot Generalization for CLIP with Synthesized Promptscited by 51
  25. 2021VT-CLIP: Enhancing Vision-Language Models with Visual-guided Textscited by 42
  26. 2021Few-Shot Object Detection via Association and DIscriminationcited by 39
  27. 2022SgVA-CLIP: Semantic-guided Visual Adapting of Vision-Language Models for Few-shot Image Classificationcited by 34
  28. 2024Transductive Zero-Shot and Few-Shot CLIPcited by 32
  29. 2023Enhancing CLIP with CLIP: Exploring Pseudolabeling for Limited-Label Prompt Tuningcited by 30
  30. 2021Few-Shot Object Detection: A Comprehensive Surveycited by 29
  31. 2023Intra-Modal Proxy Learning for Zero-Shot Visual Categorization with CLIPcited by 27
  32. 2024AMU-Tuning: Effective Logit Bias for CLIP-based Few-shot Learningcited by 22
  33. 2023Domain Aligned CLIP for Few-shot Classificationcited by 16
  34. 2023Proto-CLIP: Vision-Language Prototypical Network for Few-Shot Learningcited by 12
  35. 2024Data-Free Generalized Zero-Shot Learningcited by 8
  36. 2024Fully Fine-tuned CLIP Models are Efficient Few-Shot Learnerscited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 4; 2026: 7.

11 papers.

  1. 2026Reclaiming Lost Text Layers for Source-Free Cross-Domain Few-Shot Learningcited by 5
  2. 2026Mind the Discriminability Trap in Source-Free Cross-domain Few-shot Learningcited by 4
  3. 2026Interpretable Cross-Domain Few-Shot Learning with Rectified Target-Domain Local Alignmentcited by 4
  4. 2025SeMoBridge: Semantic Modality Bridge for Efficient Few-Shot Adaptation of CLIPcited by 3
  5. 2025Video-based Generalized Category Discovery via Memory-Guided Consistency-Aware Contrastive Learningcited by 1
  6. 2026Reevaluating the Intra-Modal Misalignment Hypothesis in CLIPcited by 1
  7. 2025Sculpting Margin Penalty: Intra-Task Adapter Merging and Classifier Calibration for Few-Shot Class-Incremental Learningcited by 0
  8. 2025Preserve and Sculpt: Manifold-Aligned Fine-tuning of Vision-Language Models for Few-Shot Learningcited by 0
  9. 2026OmniVaT: Single Domain Generalization for Multimodal Visual-Tactile Learningcited by 0
  10. 2026TCLA: Training-Free Class-wise Logit Adaptation for Medical Vision-Language Modelscited by 0
  11. 2026Reprogramming Vision-Language Models via Structured Prompt Reparameterizationcited by 0