arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.00622stat.MLcs.CVcs.LG
收藏

数据模型:从训练数据预测预测

Datamodels: Predicting Predictions from Training Data

MIT(麻省理工学院(MIT))

Andrew Ilyas, Sung Min Park, Logan Engstrom, Guillaume Leclerc, Aleksander Madry

摘要:

我们提出了一个概念框架,即数据建模,用于根据训练数据分析模型类的行为。对于任何固定的“目标”示例$x$、训练集$S$和学习算法,数据模型是一个参数化函数$2^S \to \mathbb{R}$,它对于$S' \subset S$的任何子集--仅使用关于$S'$中包含哪些$S$示例的信息--预测在$S'$上训练模型和在$x$上评估的结果。尽管被近似的基础过程的潜在复杂性(例如,深度神经网络的端到端训练和评估),我们证明了即使是简单的线性数据模型也可以成功预测模型输出。然后,我们证明了数据模型会产生各种应用,例如:准确预测数据集反事实的影响;识别脆弱的预测;找到语义相似的例子;量化训练测试泄漏;以及将数据嵌入到行为良好且功能丰富的表示空间中。本文的数据(包括预先计算的数据模型以及来自400万个经过训练的深度神经网络的原始预测)可在https://github.com/MadryLab/datamodels-data上获得。

英文摘要:

We present a conceptual framework, datamodeling, for analyzing the behavior of a model class in terms of the training data. For any fixed "target" example $x$, training set $S$, and learning algorithm, a datamodel is a parameterized function $2^S \to \mathbb{R}$ that for any subset of $S' \subset S$ -- using only information about which examples of $S$ are contained in $S'$ -- predicts the outcome of training a model on $S'$ and evaluating on $x$. Despite the potential complexity of the underlying process being approximated (e.g., end-to-end training and evaluation of deep neural networks), we show that even simple linear datamodels can successfully predict model outputs. We then demonstrate that datamodels give rise to a variety of applications, such as: accurately predicting the effect of dataset counterfactuals; identifying brittle predictions; finding semantically similar examples; quantifying train-test leakage; and embedding data into a well-behaved and feature-rich representation space. Data for this paper (including pre-computed datamodels as well as raw predictions from four million trained deep neural networks) is available at https://github.com/MadryLab/datamodels-data .

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。

↑