Pre-trained Encoders for Global Child Development: Transfer Learning Enables Deployment in Data-Scarce Settings

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Fahim, Md Muhtasim Munif, Karim, Md Rezaul
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911405322534912
author Fahim, Md Muhtasim Munif
Karim, Md Rezaul
author_facet Fahim, Md Muhtasim Munif
Karim, Md Rezaul
contents A large number of children experience preventable developmental delays each year, yet the deployment of machine learning in new countries has been stymied by a data bottleneck: reliable models require thousands of samples, while new programs begin with fewer than 100. We introduce the first pre-trained encoder for global child development, trained on 357,709 children across 44 countries using UNICEF survey data. With only 50 training samples, the pre-trained encoder achieves an average AUC of 0.65 (95% CI: 0.56-0.72), outperforming cold-start gradient boosting at 0.61 by 8-12% across regions. At N=500, the encoder achieves an AUC of 0.73. Zero-shot deployment to unseen countries achieves AUCs up to 0.84. We apply a transfer learning bound to explain why pre-training diversity enables few-shot generalization. These results establish that pre-trained encoders can transform the feasibility of ML for SDG 4.2.1 monitoring in resource-constrained settings.
format Preprint
id arxiv_https___arxiv_org_abs_2601_20987
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Pre-trained Encoders for Global Child Development: Transfer Learning Enables Deployment in Data-Scarce Settings
Fahim, Md Muhtasim Munif
Karim, Md Rezaul
Machine Learning
A large number of children experience preventable developmental delays each year, yet the deployment of machine learning in new countries has been stymied by a data bottleneck: reliable models require thousands of samples, while new programs begin with fewer than 100. We introduce the first pre-trained encoder for global child development, trained on 357,709 children across 44 countries using UNICEF survey data. With only 50 training samples, the pre-trained encoder achieves an average AUC of 0.65 (95% CI: 0.56-0.72), outperforming cold-start gradient boosting at 0.61 by 8-12% across regions. At N=500, the encoder achieves an AUC of 0.73. Zero-shot deployment to unseen countries achieves AUCs up to 0.84. We apply a transfer learning bound to explain why pre-training diversity enables few-shot generalization. These results establish that pre-trained encoders can transform the feasibility of ML for SDG 4.2.1 monitoring in resource-constrained settings.
title Pre-trained Encoders for Global Child Development: Transfer Learning Enables Deployment in Data-Scarce Settings
topic Machine Learning
url https://arxiv.org/abs/2601.20987