Research map / Label-efficient and robust learning / Recognition and long-tail
ESMC: MLLM-Based Embedding Selection for Explainable Multiple Clustering
Xinyue Wang, Yuheng Jia, Hui Liu, Junhui Hou
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 MLP clustering head with pseudo-labels (ablation) | ESMC · CMU_face Pose | NMI: 0.5498 -> 0.6271 | Improved |
| add MLP clustering head with pseudo-labels (ablation) | ESMC · Fruit360 Species | NMI: 0.3806 -> 0.5036 | Improved |
| add MLP clustering head with pseudo-labels (ablation) | ESMC · CIFAR-10 Type | NMI: 0.7149 -> 0.8293 | Improved |
| add MLP clustering head with pseudo-labels (ablation) | ESMC · Fruit Color | NMI: 0.9308 -> 0.9308 | No clear effect |
| use hidden-state token embeddings instead of MLLM text output (ablation) | ESMC w/o MLP · CIFAR-10 Environment | NMI: 0.0443 -> 0.5484 | Improved |
| use hidden-state token embeddings instead of MLLM text output (ablation) | ESMC w/o MLP · CMU_face Pose | NMI: 0.0913 -> 0.5498 | Improved |
| use hidden-state token embeddings instead of MLLM text output (ablation) | ESMC w/o MLP · CIFAR-10 Type | NMI: 0.7505 -> 0.7149 | Hurt |
| use ESMC instead of Multi-Sub | ESMC · CIFAR-10 Type | NMI: 0.5271 -> 0.8293 | Improved |
| use ESMC instead of Multi-Sub | ESMC · Fruit360 Species | NMI: 0.6123 -> 0.5036 | Hurt |
| use ESMC instead of Multi-Sub | ESMC · CMU_face Emotion | RI: 0.8527 -> 0.6697 | Hurt |
| rephrase the prompt (ablation) | ESMC · Stanford_Cars Color | NMI: 0.8138 vs 0.8016 / 0.8143 / 0.8113 | No clear effect |
| change keyword-generating LLM (GPT-4, DeepSeek-R1, Gemini-2.5) (ablation) | ESMC · Stanford_Cars | NMI: 0.8138 and 0.8817 for all | No clear effect |
Later papers that compared with this method
1 later accepted paper reports ESM-C 300M as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Greater than the Sum of Its Parts: Building Substructure into Protein Encoding Models (ICLR 2026 · better on 5 of 7 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
- Multi-Modal Proxy Learning Towards Personalized Visual Multiple Clustering (CVPR 2024 · similarity 0.87)
- Language-Assisted Image Clustering Guided by Discriminative Relational Signals and Adaptive Semantic Centers (NeurIPS 2026 · similarity 0.87)
- Clustering Properties of Self-Supervised Learning (ICML 2025 · similarity 0.86)
- Customized Multiple Clustering via Multi-Modal Subspace Proxy Learning (NeurIPS 2024 · similarity 0.86)
- Repurposing Language Models into Embedding Models: Finding the Compute-Optimal Recipe (NeurIPS 2024 · similarity 0.86)
- Image Clustering via the Principle of Rate Reduction in the Age of Pretrained Models (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.
