Research map / Label-efficient and robust learning / Recognition and long-tail
Active Learning-Based Species Range Estimation
Christian Lange, Elijah Cole, Grant Van Horn, Oisin Mac Aodha
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 |
|---|---|---|---|
| replace random sampling with hypothesis-space (HSS) active selection (ablation) | WA+ weighted-combination model · IUCN, time step 50 | MAP: 0.64 -> 0.77 | Improved |
| add online linear model h_online to hypothesis set (ablation) | WA_HSS · IUCN, time step 50 | MAP: 0.73 -> 0.77 | Improved |
| replace uncertainty sampling with HSS selection (ablation) | WA+ models · IUCN, time step 50 | MAP: 0.73 -> 0.77 | Improved |
| weighted combination of candidate models instead of logistic regression, random sampling (ablation) | random sampling · IUCN, time step 5 | MAP: 0.48 -> 0.57 | Improved |
| apply HSS selection to plain logistic regression instead of uncertainty sampling (ablation) | logistic regression · IUCN, time step 50 | MAP: 0.74 -> 0.39 | Hurt |
| use WA instead of LR with uncertainty sampling (ablation) | uncertainty sampling · IUCN, step 5 vs step 50 | MAP: 0.48 -> 0.60 at step 5; 0.74 -> 0.70 at step 50 | Mixed |
| use feature extractor deep features instead of raw location or environment features (ablation) | WA_HSS+ · IUCN | MAP | Improved |
| use randomly initialized network features instead of low-dimensional loc/env features (ablation) | WA_HSS+ · IUCN | MAP | Improved |
| reduce number of species used for backbone and candidate set (ablation) | WA_HSS · IUCN | MAP | Hurt |
| remove hypothesis likelihood weighting (ablation) | WA_HSS+ · IUCN | MAP | Hurt |
| average hard votes instead of soft votes (ablation) | WA_HSS+ · IUCN | MAP | No clear effect |
| add false-negative observation noise up to 50% (ablation) | WA_HSS+ · IUCN | MAP | Mixed |
Closest papers by meaning
- Feedforward Few-shot Species Range Estimation (ICML 2025 · similarity 0.90)
- Combining Observational Data and Language for Species Range Estimation (NeurIPS 2024 · similarity 0.90)
- RANGE: Retrieval Augmented Neural Fields for Multi-Resolution Geo-Embeddings (CVPR 2025 · similarity 0.86)
- Active Target Discovery under Uninformative Priors: The Power of Permanent and Transient Memory (NeurIPS 2025 · similarity 0.84)
- GeoPlant: Spatial Plant Species Prediction Dataset (NeurIPS 2024 · similarity 0.84)
- Open-Insect: Benchmarking Open-Set Recognition of Novel Species in Biodiversity Monitoring (NeurIPS 2025 · 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.
