通过建模模型计算分解并编辑预测结果
Decomposing and Editing Predictions by Modeling Model Computation
AI总结:
该研究提出组件建模任务及可扩展算法COAR以估计模型组件对预测的反事实影响,并验证其可支撑多类模型编辑任务。
摘要:
机器学习模型的内部计算如何将输入转化为预测?本文提出了一项名为component modeling(组件建模)的任务以解答该问题。组件建模的目标是依据模型的组成组件——即构成模型计算“基石”的简单函数(如卷积滤波器、注意力头)——来分解机器学习模型的预测结果。我们聚焦于该任务的一个特例:component attribution(组件归因),其目标是估计单个组件对给定预测的反事实影响。随后我们提出COAR,一种可扩展的组件归因估计算法;我们在不同模型、数据集和模态上验证了其有效性。最后,我们表明,用COAR估计的组件归因可直接支持五项任务上的模型编辑,分别为:修复模型错误、“遗忘”特定类别、提升子群体鲁棒性、定位后门攻击,以及增强对排版攻击的鲁棒性。我们在https://github.com/MadryLab/modelcomponents 提供了COAR的代码。
英文摘要:
How does the internal computation of a machine learning model transform inputs into predictions? In this paper, we introduce a task called component modeling that aims to address this question. The goal of component modeling is to decompose an ML model's prediction in terms of its components -- simple functions (e.g., convolution filters, attention heads) that are the "building blocks" of model computation. We focus on a special case of this task, component attribution, where the goal is to estimate the counterfactual impact of individual components on a given prediction. We then present COAR, a scalable algorithm for estimating component attributions; we demonstrate its effectiveness across models, datasets, and modalities. Finally, we show that component attributions estimated with COAR directly enable model editing across five tasks, namely: fixing model errors, ``forgetting'' specific classes, boosting subpopulation robustness, localizing backdoor attacks, and improving robustness to typographic attacks. We provide code for COAR at https://github.com/MadryLab/modelcomponents .
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。