用影响函数解析元学习中任务的贡献，让模型适应过程更透明。
TLXML: Task-Level Explanation of Meta-Learning via Influence Functions
- 将影响函数扩展到元学习双层结构，量化每项训练任务的影响
- 计算复杂度从 $O(pq^2)$ 降至 $O(pq)$，提升可扩展性
- 适用于平坦损失曲面，适合关注模型解释性的研究者
元学习使模型能快速适应新任务，但其适应机制仍不透明，尤其不清楚过往训练任务如何影响未来预测。本文提出 TLXML（任务级元学习解释框架），将影响函数拓展至元学习场景，提供适应与推理阶段的任务级解释。通过重构元学习双层结构的影响函数，量化每个元训练任务对适配后模型行为的贡献。为提升可扩展性，提出基于高斯-牛顿的近似方法，将计算复杂度从 $O(pq^2)$ 降低至 $O(pq)$，其中 $p$ 和 $q$ 分别代表模型和元参数数量。此外，引入使用伪逆海森矩阵定义的广义影响函数，可在损失曲面存在平坦方向时依然适用。实验表明，TLXML能有效按任务对下游性能的影响程度排序，生成简洁直观、符合用户抽象层次的解释。本工作为构建可解释且可信的元学习系统迈出关键一步。
原文摘要 · Abstract (English)
Meta-learning enables models to rapidly adapt to new tasks by leveraging prior experience, but its adaptation mechanisms remain opaque, especially regarding how past training tasks influence future predictions. We introduce TLXML (Task-Level eXplanation of Meta-Learning), a novel framework that extends influence functions to meta-learning settings and provides task-level explanations of adaptation and inference. By reformulating influence functions for the bi-level structure of meta-learning, we quantify the contribution of each meta-training task to the adapted model's behaviour. To ensure scalability, we propose a Gauss-Newton-based approximation that significantly reduces computational complexity from $O(pq^2)$ to $O(pq)$, where $p$ and $q$ denote the numbers of model and meta parameters, respectively. Moreover, we propose generalized influence functions defined using pseudo-inverse Hessian, which are applicable even when the loss landscape has flat directions. Results demonstrate that TLXML effectively ranks training tasks by their influence on downstream performance, offering concise, intuitive explanations aligned with user-level abstraction. This work provides a critical step toward interpretable and trustworthy meta-learning systems.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。