Neural Processes with Normalizing Flows for Wheat Height Estimation
- 1ETH Zuerich
Abstract
In this work, we investigate modeling plant traits over time using neural processes, a class of machine learning models that learn distributions over functions. Plant growth is an inherently stochastic process with complex dynamics measured mostly at irregular times throughout the growing seasons. While individual trait trajectories may be simple, their distributions are shaped by complex interactions between genotype, environment, and other factors. In particular, we focus on plant height in wheat, a deceptively simple-looking trait with complex dynamics. To model these trajectory distributions, we evaluate neural processes and in particular extensions using normalizing flows, with different combinations of genotype and environmental covariates. For controlled evaluations, we generate synthetic wheat height trajectories calibrated against Swiss weather station records and the FIP1 dataset. To fully evaluate these trajectory distributions, we use signatures, vector representations of sequential data, together with Sig-MMD and the recently introduced CSig-MMD. Sig-MMD enables direct pathwise comparison of predicted and simulator trajectory distributions, while CSig-MMD focuses this comparison on the tail, including lodged trajectories. Together, these metrics allow us to assess whether the models capture the full distribution of growth trajectories, including rare outcomes.
Related
- Vision normalizing flows for the probability-informed detection of banana diseases from in-field images
- Assessing UAV imagery and high-resolution LiDAR for tree height estimation: The role of flight speed and image overlap
- Saccadic Neural Command Estimation via Forward Dynamics Control: Influence of Extraocular Muscle Geometry on Physiological Plausibility
- Neural dynamics of auditory attention: Interactions between top-down and bottom-up processes
- Sensory representation by non-equilibrium flows of neural activity at the mesoscale of cortex
BibTeX
@misc{boss2026neural,
title = {{Neural Processes with Normalizing Flows for Wheat Height Estimation}},
author = {M. Boss and M. Volpi and L. Roth},
howpublished = {bioRxiv},
year = {2026},
month = jun,
doi = {10.64898/2026.06.24.734247},
url = {https://www.biorxiv.org/content/10.64898/2026.06.24.734247v2},
note = {CC BY},
}
RIS
TY - UNPB
TI - Neural Processes with Normalizing Flows for Wheat Height Estimation
AU - M. Boss
AU - M. Volpi
AU - L. Roth
PY - 2026
DA - 2026/06/24
PB - bioRxiv
DO - 10.64898/2026.06.24.734247
UR - https://www.biorxiv.org/content/10.64898/2026.06.24.734247v2
AB - In this work, we investigate modeling plant traits over time using neural processes, a class of machine learning models that learn distributions over functions. Plant growth is an inherently stochastic process with complex dynamics measured mostly at irregular times throughout the growing seasons. While individual trait trajectories may be simple, their distributions are shaped by complex interactions between genotype, environment, and other factors. In particular, we focus on plant height in wheat, a deceptively simple-looking trait with complex dynamics. To model these trajectory distributions, we evaluate neural processes and in particular extensions using normalizing flows, with different combinations of genotype and environmental covariates. For controlled evaluations, we generate synthetic wheat height trajectories calibrated against Swiss weather station records and the FIP1 dataset. To fully evaluate these trajectory distributions, we use signatures, vector representations of sequential data, together with Sig-MMD and the recently introduced CSig-MMD. Sig-MMD enables direct pathwise comparison of predicted and simulator trajectory distributions, while CSig-MMD focuses this comparison on the tail, including lodged trajectories. Together, these metrics allow us to assess whether the models capture the full distribution of growth trajectories, including rare outcomes.
ER -