Spatial Implicit Neural Representations for Global-Scale Species Mapping

Elijah Cole, Grant Van Horn, Christian Lange, Alexander Shepard, Patrick Leary, Pietro Perona, Scott Loarie, Oisin Mac Aodha
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:6320-6342, 2023.

Abstract

Estimating the geographical range of a species from sparse observations is a challenging and important geospatial prediction problem. Given a set of locations where a species has been observed, the goal is to build a model to predict whether the species is present or absent at any location. This problem has a long history in ecology, but traditional methods struggle to take advantage of emerging large-scale crowdsourced datasets which can include tens of millions of records for hundreds of thousands of species. In this work, we use Spatial Implicit Neural Representations (SINRs) to jointly estimate the geographical range of 47k species simultaneously. We find that our approach scales gracefully, making increasingly better predictions as we increase the number of species and the amount of data per species when training. To make this problem accessible to machine learning researchers, we provide four new benchmarks that measure different aspects of species range estimation and spatial representation learning. Using these benchmarks, we demonstrate that noisy and biased crowdsourced data can be combined with implicit neural representations to approximate expert-developed range maps for many species.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-cole23a, title = {Spatial Implicit Neural Representations for Global-Scale Species Mapping}, author = {Cole, Elijah and Horn, Grant Van and Lange, Christian and Shepard, Alexander and Leary, Patrick and Perona, Pietro and Loarie, Scott and Mac Aodha, Oisin}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {6320--6342}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/cole23a/cole23a.pdf}, url = {https://proceedings.mlr.press/v202/cole23a.html}, abstract = {Estimating the geographical range of a species from sparse observations is a challenging and important geospatial prediction problem. Given a set of locations where a species has been observed, the goal is to build a model to predict whether the species is present or absent at any location. This problem has a long history in ecology, but traditional methods struggle to take advantage of emerging large-scale crowdsourced datasets which can include tens of millions of records for hundreds of thousands of species. In this work, we use Spatial Implicit Neural Representations (SINRs) to jointly estimate the geographical range of 47k species simultaneously. We find that our approach scales gracefully, making increasingly better predictions as we increase the number of species and the amount of data per species when training. To make this problem accessible to machine learning researchers, we provide four new benchmarks that measure different aspects of species range estimation and spatial representation learning. Using these benchmarks, we demonstrate that noisy and biased crowdsourced data can be combined with implicit neural representations to approximate expert-developed range maps for many species.} }
Endnote
%0 Conference Paper %T Spatial Implicit Neural Representations for Global-Scale Species Mapping %A Elijah Cole %A Grant Van Horn %A Christian Lange %A Alexander Shepard %A Patrick Leary %A Pietro Perona %A Scott Loarie %A Oisin Mac Aodha %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-cole23a %I PMLR %P 6320--6342 %U https://proceedings.mlr.press/v202/cole23a.html %V 202 %X Estimating the geographical range of a species from sparse observations is a challenging and important geospatial prediction problem. Given a set of locations where a species has been observed, the goal is to build a model to predict whether the species is present or absent at any location. This problem has a long history in ecology, but traditional methods struggle to take advantage of emerging large-scale crowdsourced datasets which can include tens of millions of records for hundreds of thousands of species. In this work, we use Spatial Implicit Neural Representations (SINRs) to jointly estimate the geographical range of 47k species simultaneously. We find that our approach scales gracefully, making increasingly better predictions as we increase the number of species and the amount of data per species when training. To make this problem accessible to machine learning researchers, we provide four new benchmarks that measure different aspects of species range estimation and spatial representation learning. Using these benchmarks, we demonstrate that noisy and biased crowdsourced data can be combined with implicit neural representations to approximate expert-developed range maps for many species.
APA
Cole, E., Horn, G.V., Lange, C., Shepard, A., Leary, P., Perona, P., Loarie, S. & Mac Aodha, O.. (2023). Spatial Implicit Neural Representations for Global-Scale Species Mapping. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:6320-6342 Available from https://proceedings.mlr.press/v202/cole23a.html.

Related Material