Research map / Label-efficient and robust learning / Self-supervised and contrastive
HASSL: Hierarchy-Aware Self-Supervised Learning Framework for Single Cell Microscopy
Julius Riel, Vishwa Mohan Singh, SAI ANIRUDH ARYASOMAYAJULA, Anuun Chinbat, Hannes Leonhard, Moritz Ladenburger, Frederik Alexander, Vishisht Choudhary, Fabio Laredo, Giacomo Masserdotti, Thorben Prein, Amirhossein Kardoost and 1 more
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 double-teacher distillation and stability-weighted HDBSCAN loss | DINOv3 · Single-cell retrieval | Acc@9: 80.9 -> 83.5 | Improved |
| add HASSL to DINOv3 | DINOv3 · Multi-level hierarchy datasets | top-9 retrieval: +6.3% | Improved |
| add HASSL | DINOv3 · Drug classification (Allen perturbation) | F1 weighted: 81.8 -> 92.0 | Improved |
| use flat DBSCAN instead of HDBSCAN (ablation) | DINOv3 · Single-cell retrieval | Acc@9: 80.9 -> 75.6 | Hurt |
| remove stability weighting from HDBSCAN loss (ablation) | DINOv3 · Single-cell retrieval | Acc@9: 80.9 -> 74.5 | Hurt |
| add double-teacher distillation (ablation) | HASSL · Single-cell retrieval | Acc@9: 82.8 -> 83.5 | Improved |
| add HDBSCAN loss (ablation) | HASSL · Single-cell retrieval | mAP: 57.4 -> 57.8 | Mixed |
| add double-teacher distillation (ablation) | HASSL · Human Protein Atlas | accuracy: 50.1 -> 54.2 | Improved |
| add HASSL vs domain-specific Cellpaint-DINO | HASSL · Human Protein Atlas | accuracy: 54.8 vs 54.2 | Hurt |
| add HASSL | DINOv3 · Curated dataset classification | accuracy: 44.9 -> 45.6 | Improved |
| add HASSL | DINOv3 · AICS cross-modal retrieval | Acc@9: 77.6 -> 92.2 | Improved |
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.
- HCSC — HCSC: Hierarchical Contrastive Selective Coding (CVPR 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
- ViTally Consistent: Scaling Biological Representation Learning for Cell Microscopy (ICML 2025 · similarity 0.87)
- Towards Scalable Context-Aware Single-Cell Spatial Transcriptomics Prediction from Histology Images (NeurIPS 2026 · similarity 0.86)
- Masked Autoencoders for Microscopy are Scalable Learners of Cellular Biology (CVPR 2024 · similarity 0.86)
- Micro-Bench: A Microscopy Benchmark for Vision-Language Understanding (NeurIPS 2024 · similarity 0.85)
- MorphoHELM: A Comprehensive Benchmark for Evaluating Representations for Microscopy-Based Morphology Assays (NeurIPS 2026 · similarity 0.85)
- $\epsilon$-Seg: Sparsely Supervised Semantic Segmentation of Microscopy Data (NeurIPS 2025 · 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.
