Research map / Vision-language and multimodal / Human and embodied modeling
Agile Deliberation: Concept Deliberation for Subjective Visual Classification
Leijie Wang, Otilia Stretcu, Wei Qiao, Thomas Denby, Krishnamurthy Viswanathan, Enming Luo, Chun-Ta Lu, Tushar Dogra, Ranjay Krishna, Ariel Fuxman
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 |
|---|---|---|---|
| add human-in-the-loop concept deliberation (Agile Deliberation) over zero-shot VLM classifier | Gemini 2.5 Flash classifier · Paid to play | F1: 0.48 -> 0.59 | Improved |
| add human-in-the-loop concept deliberation (Agile Deliberation) over zero-shot VLM classifier | Gemini 2.5 Flash classifier · Healthy food | F1: 0.48 -> 0.58 | Improved |
| automatic LLM concept decomposition without user feedback (Modeling Collaborator) | zero-shot VLM classifier · Paid to play | F1: 0.48 -> 0.53 | Improved |
| automatic LLM concept decomposition without user feedback (Modeling Collaborator) | zero-shot VLM classifier · Healthy food (Manual participants) | F1: 0.82 -> 0.78 | Hurt |
| manual deliberation with image search and prompt editing | zero-shot VLM classifier · Healthy food (Manual participants) | F1: 0.82 -> 0.79 | Hurt |
| manual deliberation with image search and prompt editing | zero-shot VLM classifier · Paid to play (Manual participants) | F1: 0.43 -> 0.51 | Improved |
| Agile Deliberation versus zero-shot, precision and recall | Agile Deliberation · Paid to play | Precision: 0.35 -> 0.57 (recall 0.80 -> 0.63) | Improved |
| Agile Deliberation instead of manual deliberation | user study participants · User survey | Effort required (1-7, lower better): 4.67 -> 3.11 | Improved |
| Agile Deliberation instead of manual deliberation | user study participants · User survey | Feeling insecure, stressed or annoyed: 3.00 -> 1.67 | Improved |
| Agile Deliberation instead of manual deliberation | user study participants · User survey | Mental demand: 4.56 -> 3.22 | No clear effect |
| Agile Deliberation instead of manual deliberation | user study participants · User survey | Success in articulating concept ideas: 5.11 -> 5.56 | No clear effect |
| more deliberation rounds | Agile Deliberation · Paid to play and Healthy food | F1 | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Modeling Collaborator — Modeling Collaborator: Enabling Subjective Vision Classification With Minimal Human Effort via LLM Tool-Use (CVPR 2024) · better on 2 of 2 shared results
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
- Modeling Collaborator: Enabling Subjective Vision Classification With Minimal Human Effort via LLM Tool-Use (CVPR 2024 · similarity 0.89)
- ConceptBed: Evaluating Concept Learning Abilities of Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.85)
- UniBench: Visual Reasoning Requires Rethinking Vision-Language Beyond Scaling (NeurIPS 2024 · similarity 0.85)
- ConceptScope: Characterizing Dataset Bias via Disentangled Visual Concepts (NeurIPS 2025 · similarity 0.84)
- Automated Detection of Visual Attribute Reliance with a Self-Reflective Agent (NeurIPS 2025 · similarity 0.84)
- Probabilistic Conceptual Explainers: Trustworthy Conceptual Explanations for Vision Foundation Models (ICML 2024 · similarity 0.84)
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.
