paper-with-me

홈 › Papers

Source-Optimal Training is Transfer-Suboptimal

2025-11-11 · C. Evans Hedges arxiv

We prove that training a source model optimally for its own task is generically suboptimal when the objective is downstream transfer. We study the source-side optimization problem in L2-SP ridge regression and show a fundamental mismatch between the source-optimal and transfer-optimal source regularization: outside of a measure-zero set, $τ_0^* \neq τ_S^*$. We characterize the transfer-optimal source penalty $τ_0^*$ as a function of task alignment and identify an alignment-dependent reversal: with imperfect alignment ($0<ρ<1$), transfer benefits from stronger source regularization, while in super-aligned regimes ($ρ>1$), transfer benefits from weaker regularization. Additionally, in isotropic settings, the decision of whether transfer helps is independent of the target sample size and noise, depending only on task alignment and source characteristics. We verify the linear predictions in a synthetic ridge regression experiment, and we present experiments on MNIST, CIFAR-10, and 20 Newsgroups as evidence that the source-optimal versus transfer-optimal mismatch persists in standard nonlinear transfer learning pipelines.

📄 PDF Abstract BibTeX arXiv:2511.08401

Code (0)

등록된 구현이 없습니다.

Tasks

Transfer Learning

Similar Papers 제목 키워드 기반

Robust Fine-Tuning from Non-Robust Pretrained Models: Mitigating Suboptimal Transfer With Epsilon-Scheduling

2025-09-27 · Jonas Ngnawé, Maxime Heuillet, Sabyasachi Sahoo, Yann Pequignot 외 arxiv

Fine-tuning pretrained models is a standard and effective workflow in modern machine learning. However, robust fine-tuning (RFT), which aims to simultaneously achieve adaptation to a downstream task and robustness to adv…

A Theoretical Framework for Data Efficient Multi-Source Transfer Learning Based on Cramér-Rao Bound

2025-02-06 · Qingyue Zhang, Haohao Fu, Guanbo Huang, Yaoyuan Liang 외

Multi-source transfer learning provides an effective solution to data scarcity in real-world supervised learning scenarios by leveraging multiple source tasks. In this field, existing works typically use all available sa…

Transfer Learning

Filtering Learning Histories Enhances In-Context Reinforcement Learning

2025-05-21 · Weiqin Chen, Xinjie Zhang, Dharmashankar Subramanian, Santiago Paternain

Transformer models (TMs) have exhibited remarkable in-context reinforcement learning (ICRL) capabilities, allowing them to generalize to and improve in previously unseen environments without re-training or fine-tuning. T…

In-Context Reinforcement Learningreinforcement-learningReinforcement Learning

Meta-learning Transferable Representations with a Single Target Domain

2020-11-03 · Hong Liu, Jeff Z. HaoChen, Colin Wei, Tengyu Ma

Recent works found that fine-tuning and joint training---two popular approaches for transfer learning---do not always improve accuracy on downstream tasks. First, we aim to understand more about when and why fine-tuning …

Meta-LearningRepresentation LearningTransfer Learning

Ambient Diffusion Policy: Imitation Learning from Suboptimal Data in Robotics

2026-06-10 · Adam Wei, Nicholas Pfaff, Thomas Cohn, Arif Kerem Dayı 외 arxiv

We propose Ambient Diffusion Policy, a simple and principled method for imitation learning from suboptimal data in robotics. High-quality, task-specific robot data is expensive and time-consuming to collect, while subopt…