Research map / Vision-language and multimodal / Multimodal LLMs
VLSlice: Interactive Vision-and-Language Slice Discovery
Eric Slyman, Minsuk Kahng, Stefan Lee
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 VLSlice (clusters, similar/counterfactual refinement, correlation plot) instead of ListSort | slice discovery interface · user study, both tasks | images cataloged: +84.58 images (p=0.017) | Improved |
| use VLSlice instead of ListSort | slice discovery interface · user study, both tasks | number of slices: +0.5955 slices (p=0.295) | No clear effect |
| use VLSlice instead of ListSort | slice discovery interface · user study, both tasks | slice coherency F1: +0.215 F1 (p=0.006) | Improved |
| use VLSlice instead of ListSort | slice discovery interface · user study, both tasks | missed images (representativeness): -2.1 missed (p=0.35) | No clear effect |
| use VLSlice instead of ListSort | slice discovery interface · House/Nice task | missed images: 41 -> 70 | Hurt |
| use VLSlice instead of ListSort | slice discovery interface · post-study Likert ratings | enjoyment / would use again: 5.27 -> 6.45 (enjoyed); 4.73 -> 6.45 (use again) | Improved |
| use VLSlice instead of ListSort | slice discovery interface · post-study Likert ratings | ease of learning / ease of use: 6.27 -> 5.55 (learn); 6.00 -> 5.55 (use) | Hurt |
| use VLSlice instead of ListSort | slice discovery interface · Person/CEO slices | share of slices on gender, skin tone, age: 34% -> 62% | Improved |
| use VLSlice instead of ListSort | slice discovery interface · House/Nice slices | share of low-level color/structure/photographic slices: 45% -> 14% | Improved |
| use change in augmented caption percentile instead of raw augmented caption similarity | bias measure · extended captions (qualitative) | effect of canonical images | Improved |
Closest papers by meaning
- Calibrating Multi-modal Representations: A Pursuit of Group Robustness without Annotations (CVPR 2024 · similarity 0.89)
- Beyond the Linear Separability Ceiling: Aligning Representations in VLMs (NeurIPS 2026 · similarity 0.88)
- LVLM-Aided Alignment of Task-Specific Vision Models (CVPR 2026 · similarity 0.87)
- Discover and Mitigate Multiple Biased Subgroups in Image Classifiers (CVPR 2024 · similarity 0.87)
- Synthesize Diagnose and Optimize: Towards Fine-Grained Vision-Language Understanding (CVPR 2024 · similarity 0.86)
- Mordal: Automated Pretrained Model Selection for Vision Language Models (ICLR 2026 · similarity 0.86)
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.
