mlsanityData to 14 Oct 2026

Language-Aware Information Maximization for Transductive Few-Shot CLIP

arXiv 2509.00305 ↗First posted 30 Aug 2025cs.CVcited by 0

On the map: Multimodal LLMs / vision-language models → Prompt learning and adaptation of CLIP

Abstract (arXiv)

Transductive few-shot learning has triggered an abundant literature focusing on vision-only models, but is still at a nascent stage within the recent context of foundational vision-language models (VLMs). Only a few recent methods addressed the problem, pointing to the potential of tranduction in VLMs and to the need for VLM-tailored methods. Building on this momentum, we leverage information-theoretic concepts and recent progress in parameter-efficient fine-tuning (PEFT), developing a highly competitive transductive few-shot CLIP method. Specifically, we introduce a novel Language-aware Information MaximizatiOn (LIMO) loss integrating three complementary terms: (i) the mutual information between the vision inputs and the textual class descriptions; (ii) a Kullback-Leibler (KL) divergence penalizing deviation of the network's probabilistic outputs from the text-driven zero-shot predictions; and (iii) a standard cross-entropy loss based on the labeled shots. Furthermore, we challenge the commonly followed fine-tuning practices in the context of transductive few-shot learning, and explore PEFT strategies, completely overlooked in this context. Surprisingly, we observe substantial boosts in performances, which points to the potential of adapting a subset of the model's parameters in the transductive few-shot setting. We report comprehensive evaluations, which show that LIMO outperforms the very recent transductive few-shot CLIP methods by a large margin and yields significant gains over the best-performing inductive methods. Our code is publicly available at:[ hrefhttps://github.com/ghassenbaklouti/LIMOhere ]

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Prompt learning and adaptation of CLIP, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 45 of its 58 references that are in the map. First posted on arXiv 30 Aug 2025.

Builds on

45 of 58 references.

Closest ancestors · 11

Same direction, newest first

  1. 2024Boosting Vision-Language Models with Transduction3 Jun
  2. 2024Low-Rank Few-Shot Adaptation of Vision-Language Models28 May
  3. 2024On the test-time zero-shot generalization of vision-language models: Do we really need prompt learning?3 May
  4. 2024Label Propagation for Zero-shot Classification with Vision-Language Models5 Apr
  5. 2023Visual-Language Prompt Tuning with Knowledge-guided Context Optimization23 Mar
  6. 2022Test-Time Prompt Tuning for Zero-Shot Generalization in Vision-Language Models15 Sep
  7. 2022Tip-Adapter: Training-free Adaption of CLIP for Few-shot Classification19 Jul
  8. 2022Prompt-aligned Gradient for Prompt Tuning30 May
  9. 2022Conditional Prompt Learning for Vision-Language Models10 Mar
  10. 2021CLIP-Adapter: Better Vision-Language Models with Feature Adapters9 Oct
  11. 2021Learning to Prompt for Vision-Language Models2 Sep

Other references · 34

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  3. 2023QLoRA: Efficient Finetuning of Quantized LLMscited by 3,418
  4. 2013Describing Textures in the Wildcited by 2,125
  5. 2013Fine-Grained Visual Classification of Aircraftcited by 1,726
  6. 2022Visual Prompt Tuningcited by 1,681
  7. 2017EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classificationcited by 1,496
  8. 2018Learning deep representations by mutual information estimation and maximizationcited by 1,273
  9. 2021BitFit: Simple Parameter-efficient Fine-tuning for Transformer-based Masked Language-modelscited by 1,245
  10. 2023Vision-Language Models for Vision Tasks: A Surveycited by 839
  11. 2019Learning Representations by Maximizing Mutual Information Across Viewscited by 734
  12. 2022AdaptFormer: Adapting Vision Transformers for Scalable Visual Recognitioncited by 719
22 more
  1. 2022MaPLe: Multi-modal Prompt Learningcited by 687
  2. 2019A Closer Look at Few-shot Classificationcited by 684
  3. 2023AdaLoRA: Adaptive Budget Allocation for Parameter-Efficient Fine-Tuningcited by 357
  4. 2019On Mutual Information Maximization for Representation Learningcited by 316
  5. 2019A Baseline for Few-Shot Image Classificationcited by 298
  6. 2022Scaling & Shifting Your Features: A New Baseline for Efficient Model Tuningcited by 294
  7. 2022DyLoRA: Parameter Efficient Tuning of Pre-trained Models using Dynamic Search-Free Low-Rank Adaptationcited by 233
  8. 2018Learning to Propagate Labels: Transductive Propagation Network for Few-shot Learningcited by 211
  9. 2017Learning Discrete Representations via Information Maximizing Self-Augmented Trainingcited by 149
  10. 2020A unifying mutual information view of metric learning: cross-entropy vs. pairwise lossescited by 100
  11. 2019Prototype Rectification for Few-Shot Learningcited by 97
  12. 2020Laplacian Regularized Few-Shot Learningcited by 92
  13. 2023One-for-All: Generalized LoRA for Parameter-Efficient Fine-tuningcited by 85
  14. 2020Leveraging the Feature Distribution in Transfer-based Few-Shot Learningcited by 74
  15. 2024LP++: A Surprisingly Strong Linear Probe for Few-Shot CLIPcited by 60
  16. 2022Parametric Information Maximization for Generalized Category Discoverycited by 39
  17. 2018Deep clustering: On the link between discriminative models and K-meanscited by 37
  18. 2024Transductive Zero-Shot and Few-Shot CLIPcited by 32
  19. 2020Exploring Category-Agnostic Clusters for Open-Set Domain Adaptationcited by 27
  20. 2024QDyLoRA: Quantized Dynamic Low-Rank Adaptation for Efficient Large Language Model Tuningcited by 26
  21. 2023Hydra: Multi-head Low-rank Adaptation for Parameter Efficient Fine-tuningcited by 17
  22. 2024Enhancing Remote Sensing Vision-Language Models for Zero-Shot Scene Classificationcited by 8

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2023A Closer Look at the Few-Shot Adaptation of Large Vision-Language Models
  2. 2024Fully Fine-tuned CLIP Models are Efficient Few-Shot Learners
  3. 2023Robust Fine-Tuning of Vision-Language Models for Domain Generalization
  4. 2025Generate, Transduct, Adapt: Iterative Transduction with VLMs
  5. 2022Rethinking Task Sampling for Few-shot Vision-Language Transfer Learning
  6. 2024Zoom-shot: Fast and Efficient Unsupervised Zero-Shot Transfer of CLIP to Vision Encoders with Multimodal Loss
  7. 2022Localized Latent Updates for Fine-Tuning Vision-Language Models
  8. 2025Topology-Aware CLIP Few-Shot Learning