arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2502.06970cs.LGstat.ML
收藏

用于可认证少样本迁移学习的模型扩散

Weight Space Learning for Certifiable Few-shot Transfer Learning

School of Informatics, University of Edinburgh(信息学院,爱丁堡大学);Samsung AI Center, Cambridge(三星人工智能中心,剑桥)

Fady Rezk, Royson Lee, Henry Gouk, Timothy Hospedales, Minyoung Kim

AI总结:

本文提出一种基于扩散模型的少样本迁移学习方法,通过在有限PEFT参数样本集合中进行采样评估,为下游任务提供非空的学习理论泛化保证。

摘要:

在当代深度学习中,解决低数据问题的一种普遍且有效的工作流程是通过参数高效微调(PEFT)将强大的预训练基础模型(FMs)适应到新任务。然而,尽管在经验上是有效的,由此产生的解决方案缺乏泛化保证来认证其准确性——在重要应用中部署之前,可能出于伦理或法律原因需要这种保证。在本文中,我们开发了一种新颖的迁移学习方法,旨在为下游任务提供非空的学习理论泛化保证,即使在低样本机制下也是如此。具体来说,我们首先使用上游任务训练PEFT参数上的分布。然后,我们通过采样和评估程序学习下游任务——从训练好的扩散模型中采样合理的PEFT,并选择在下游数据上具有最高似然的PEFT。关键的是,这将我们的模型假设限制在有限的PEFT样本集合中。与神经网络权重的典型连续假设空间相比,这有助于获得更严格的风险证书。我们实例化了我们的界限,并展示了与现有学习方法相比非平凡的泛化保证,而现有方法在低样本机制下会导致空泛的界限。

英文摘要:

In contemporary deep learning, a prevalent and effective workflow for solving low-data problems is adapting powerful pre-trained foundation models (FMs) to new tasks via parameter-efficient fine-tuning (PEFT). However, while empirically effective, the resulting solutions lack generalisation guarantees to certify their accuracy - which may be required for ethical or legal reasons prior to deployment in high-importance applications. In this paper we develop a novel transfer learning approach that is designed to facilitate non-vacuous learning theoretic generalisation guarantees for downstream tasks, even in the low-shot regime. Specifically, we first use upstream tasks to train a distribution over PEFT parameters. We then learn the downstream task by a sample-and-evaluate procedure -- sampling plausible PEFTs from the trained diffusion model and selecting the one with the highest likelihood on the downstream data. Crucially, this confines our model hypothesis to a finite set of PEFT samples. In contrast to the typical continuous hypothesis spaces of neural network weights, this facilitates tighter risk certificates. We instantiate our bound and show non-trivial generalization guarantees compared to existing learning approaches which lead to vacuous bounds in the low-shot regime.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑