Cascaded Transfer: Learning Many Tasks under Budget Constraints

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Campagne, Eloi, Amara-Ouali, Yvenn, Goude, Yannig, Mougeot, Mathilde, Kalogeratos, Argyris
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917524299317248
author Campagne, Eloi
Amara-Ouali, Yvenn
Goude, Yannig
Mougeot, Mathilde
Kalogeratos, Argyris
author_facet Campagne, Eloi
Amara-Ouali, Yvenn
Goude, Yannig
Mougeot, Mathilde
Kalogeratos, Argyris
contents In distributed applications, such as energy demand forecasting at the substation level or federated learning, a large number of related tasks must be learned by different models, while the exact task relationships are unknown. We propose the novel Cascaded Transfer Learning (CTL) paradigm in which model parameters cascade hierarchically through tasks organized as a rooted tree, respecting a global training budget. Starting from a source task, the tree specifies the order in which tasks are learned and refined, with the budget allocated along its branches. We design cascade mechanisms based on spanning trees that connect all tasks by minimizing an objective combining pairwise task distances and the available training budget, which yield geometry-aware and depth-bounded transfer graphs. We theoretically characterize how transfer errors accumulate and attenuate along cascade paths: errors introduced at any upstream node are contracted by every downstream refinement, and balanced tree topologies bound this accumulation. Experiments on synthetic and real many-task settings, time-series forecasting and image classification, show that CTL enables more accurate and cost-effective adaptation across large task collections than alternative approaches, with the largest gains at the tightest budgets.
format Preprint
id arxiv_https___arxiv_org_abs_2601_21513
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Cascaded Transfer: Learning Many Tasks under Budget Constraints
Campagne, Eloi
Amara-Ouali, Yvenn
Goude, Yannig
Mougeot, Mathilde
Kalogeratos, Argyris
Machine Learning
In distributed applications, such as energy demand forecasting at the substation level or federated learning, a large number of related tasks must be learned by different models, while the exact task relationships are unknown. We propose the novel Cascaded Transfer Learning (CTL) paradigm in which model parameters cascade hierarchically through tasks organized as a rooted tree, respecting a global training budget. Starting from a source task, the tree specifies the order in which tasks are learned and refined, with the budget allocated along its branches. We design cascade mechanisms based on spanning trees that connect all tasks by minimizing an objective combining pairwise task distances and the available training budget, which yield geometry-aware and depth-bounded transfer graphs. We theoretically characterize how transfer errors accumulate and attenuate along cascade paths: errors introduced at any upstream node are contracted by every downstream refinement, and balanced tree topologies bound this accumulation. Experiments on synthetic and real many-task settings, time-series forecasting and image classification, show that CTL enables more accurate and cost-effective adaptation across large task collections than alternative approaches, with the largest gains at the tightest budgets.
title Cascaded Transfer: Learning Many Tasks under Budget Constraints
topic Machine Learning
url https://arxiv.org/abs/2601.21513