# InfoTaxa: Information-Calibrated Label-Free Clustering for Fine-Grained Visual Taxonomy

arXiv: https://arxiv.org/abs/2609.17218 · HTML: https://mlsanity.com/paper/2609.17218/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 18 of its 32 references that are in the map.
- First posted on arXiv 15 Sep 2026.

## Builds on (most cited first, 18 of 18)

- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [From Louvain to Leiden: guaranteeing well-connected communities](https://mlsanity.com/paper/1810.08473/) (2018)
- [EVA-02: A Visual Representation for Neon Genesis](https://mlsanity.com/paper/2303.11331/) (2023)
- [SCAN: Learning to Classify Images without Labels](https://mlsanity.com/paper/2005.12320/) (2020)
- [On Mutual Information Maximization for Representation Learning](https://mlsanity.com/paper/1907.13625/) (2019)
- [Generalized Category Discovery](https://mlsanity.com/paper/2201.02609/) (2022)
- [Formal Limitations on the Measurement of Mutual Information](https://mlsanity.com/paper/1811.04251/) (2018)
- [BioCLIP: A Vision Foundation Model for the Tree of Life](https://mlsanity.com/paper/2311.18803/) (2023)
- [Parametric Classification for Generalized Category Discovery: A Baseline Study](https://mlsanity.com/paper/2211.11727/) (2022)
- [Efficient Parameter-free Clustering Using First Neighbor Relations](https://mlsanity.com/paper/1902.11266/) (2019)
- [BioCLIP 2: Emergent Properties from Scaling Hierarchical Contrastive Learning](https://mlsanity.com/paper/2505.23883/) (2025)
- [Exploring the Limits of Deep Image Clustering using Pretrained Models](https://mlsanity.com/paper/2303.17896/) (2023)
- [BIOSCAN-5M: A Multimodal Dataset for Insect Biodiversity](https://mlsanity.com/paper/2406.12723/) (2024)
- [CLIBD: Bridging Vision and Genomics for Biodiversity Monitoring at Scale](https://mlsanity.com/paper/2405.17537/) (2024)
- [Let Go of Your Labels with Unsupervised Transfer](https://mlsanity.com/paper/2406.07236/) (2024)
- [An Empirical Study into Clustering of Unseen Datasets with Self-Supervised Encoders](https://mlsanity.com/paper/2406.02465/) (2024)

## Abstract (arXiv)

Label-free clustering of frozen pretrained visual embeddings offers a scalable route to biodiversity monitoring, but image-only fine-grained taxonomy exhibits a consistent coarse-to-fine failure mode: clusters recover broad taxonomic structure yet plateau at species level. We study this behaviour on BIOSCAN-5M through an information-calibrated clustering analysis. BioCLIP~2 features with UMAP and HDBSCAN reach 0.79 AMI at family and 0.67 at genus, substantially improving over the prior image baseline and remaining competitive with oracle-K, graph-based, and learned clustering heads on the same frozen features. To diagnose whether the remaining plateau is method-limited or information-limited, we introduce InfoTaxa, which combines clustering efficiency---the fraction of probe-estimated image information recovered by an unsupervised partition---with paired DNA as an audit signal only, not an inference input. The density pipeline recovers approximately 0.90 and 0.81 of the image-available information at order and family, respectively. Held-out late-fusion probes show that adding DNA to the image embedding reduces species-level prediction error by approximately two bits. Robustness analyses cover multiple image encoders, described-species and rare-class subsets, probe diagnostics, and held-out-species coarse-rank generalisation and same-species retrieval. Thus, in the tested setting, species-level label-free clustering is both clustering-limited and representation-limited: improved clustering may recover additional image-exposed structure, but cannot close the DNA-audited information gap alone.