Research map / Vision-language and multimodal / CLIP and zero-shot
NLIP: Noise-Robust Language-Image Pre-training
Runhui Huang, Yanxin Long, Jianhua Han, Hang Xu, Xiwen Liang, Chunjing Xu, Xiaodan Liang
This paper tried a common change: Add positional encoding
Add positional embeddings to the 75%-masked visual encoder · ImageNet zero-shot classification · top-1 accuracy (%): 26.5 -> 27.0 Improved
What 646 accepted papers report for this change
- 525 improved
- 58 no clear effect
- 7 mixed
- 56 hurt
- Learning-Augmented Scalable Linear Assignment Problem Optimization via Neural Dual Warm-Starts ICML 2026 · add positional encodings to features (d=13 -> d=21) · Block-structured, N=4096 · speedupImproved
- Mesh Based Simulations with Spatial and Temporal awareness ICML 2026 · use learned positional embedding, learnable relative bias or weighted adjacency instead of RoPE · Aneurysm · RMSEHurt
- Sparse ActionGen: Accelerating Diffusion Policy with Real-time Pruning ICML 2026 · add sinusoidal positional encoding to the pruner · Robomimic Tool Hang · success rate (%): 28 -> 54Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use noise-robust pre-training with noise harmonization and noise completion | NLIP on YFCC26M · YFCC26M zero-shot image classification, 12-dataset average, ViT-B/16 | top-1 accuracy (%): 37.3 -> 45.9 | Improved |
| Use noise-robust pre-training with noise harmonization and noise completion | NLIP on YFCC26M · Flickr30K zero-shot image-to-text retrieval | R@1: 53.9 -> 82.6 | Improved |
| Use noise-robust pre-training with noise harmonization and noise completion | NLIP on YFCC26M · MSCOCO image captioning | CIDEr: 133.3 -> 135.2 | Improved |
| Add noise completion (ablation) | NLIP · Flickr30K image-to-text retrieval | R@1: 72.2 -> 82.6 | Improved |
| Add noise completion (ablation) | NLIP · MSCOCO image captioning | CIDEr: 132.4 -> 134.0 | Improved |
| Add noise harmonization (ablation) | NLIP · Flickr30K text-to-image retrieval | R@1: 47.1 -> 61.2 | Improved |
| Add noise harmonization (ablation) | NLIP · ImageNet zero-shot classification | top-1 accuracy (%): 46.7 -> 47.4 | Improved |
| Condition noise completion on visual concepts (ablation) | NLIP · ImageNet zero-shot classification | top-1 accuracy (%): 46.7 -> 47.4 | Improved |
| Condition noise completion on visual concepts (ablation) | NLIP · Flickr30K text-to-image retrieval | R@1: 60.1 -> 61.2 | Improved |
| Use YFCC visual concepts rather than no visual concepts (ablation) | YFCC26M-CLIP · ImageNet zero-shot classification | top-1 accuracy (%): 37.3 -> 42.7 | Improved |
| Use ensemble visual concepts rather than no visual concepts (ablation) | YFCC26M-CLIP · ImageNet zero-shot classification | top-1 accuracy (%): 39.2 -> 42.5 | Improved |
| Use pre-trained BART as the text encoder (ablation) | CLIP architecture trained on YFCC26M · ImageNet zero-shot classification | top-1 accuracy (%): 30.1 -> 32.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- FILIP — FILIP: Fine-grained Interactive Language-Image Pre-Training (ICLR 2022) · better on 2 of 2 shared results
- BLIP — BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation (ICML 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Noise-Aware Image Captioning with Progressively Exploring Mismatched Words (AAAI 2024 · similarity 0.90)
- Modeling Caption Diversity in Contrastive Vision-Language Pretraining (ICML 2024 · similarity 0.88)
- Mining Fine-Grained Image-Text Alignment for Zero-Shot Captioning via Text-Only Training (AAAI 2024 · similarity 0.87)
- SmartCLIP: Modular Vision-language Alignment with Identification Guarantees (CVPR 2025 · similarity 0.87)
- SoftCLIP: Softer Cross-Modal Alignment Makes CLIP Stronger (AAAI 2024 · similarity 0.87)
- Improving Cross-Modal Alignment with Synthetic Pairs for Text-Only Image Captioning (AAAI 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
