Research map / Vision-language and multimodal / Human and embodied modeling
Make it SING: Analyzing Semantic Invariants in Classifiers
Harel Yadid, Meir Yossef Levi, Roy Betser, Guy Gilboa
Read the paperCheck your idea against 137,217 papers
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 non-linear 3-layer MLP translator instead of linear (ablation) | translator to CLIP space · Translator validation | mean cosine similarity: 0.8946 -> 0.9049 | No clear effect |
| use non-linear 4-layer MLP translator instead of linear (ablation) | translator to CLIP space · Translator validation | validation MSE: 0.091355 -> 0.084346 | No clear effect |
| use cosine-only loss instead of MSE for translator (ablation) | translator · Translator training | cosine similarity and MSE | Hurt |
| use EVA02 instead of CLIP as target space (ablation) | model-level IS/AS ratio · 5-model comparison | IS/AS ratio ordering | No clear effect |
| pretrain with self-supervised DINO ViT rather than supervised/weakly supervised CNNs | classifier · text-gradient null perturbation | |AS| toward target: 12.04 -> 5.0 (ResNet50 vs DinoViT) | Improved |
| compare ResNext101 with DinoViT on null leakage | classifier · ImageNet 1000 classes | IS/AS ratio | Hurt |
Closest papers by meaning
- Invariant Features in Language Models: Geometric Characterization and Model Attribution (NeurIPS 2026 · similarity 0.87)
- Synthesize Diagnose and Optimize: Towards Fine-Grained Vision-Language Understanding (CVPR 2024 · similarity 0.87)
- SUGARCREPE++ Dataset: Vision-Language Model Sensitivity to Semantic and Lexical Alterations (NeurIPS 2024 · similarity 0.86)
- VisMin: Visual Minimal-Change Understanding (NeurIPS 2024 · similarity 0.86)
- Not Only Text: Exploring Compositionality of Visual Representations in Vision-Language Models (CVPR 2025 · similarity 0.86)
- Follow-Up Differential Descriptions: Language Models Resolve Ambiguities for Image Classification (ICLR 2024 · similarity 0.85)
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.
