ModelPred: A Framework for Predicting Trained Model from Training Data

TL;DR

ModelPred is a neural network-based framework that predicts trained model parameters from training data, enhancing interpretability, data valuation, and model calibration in machine learning workflows.

Contribution

It introduces a novel neural set function approach with regularization techniques to directly predict model parameters from training data, differing from existing behavior-based models.

Findings

Effective in predicting model parameters across various datasets

Improves interpretability and accountability in ML workflows

Enables applications like data valuation and model calibration

Abstract

In this work, we propose ModelPred, a framework that helps to understand the impact of changes in training data on a trained model. This is critical for building trust in various stages of a machine learning pipeline: from cleaning poor-quality samples and tracking important ones to be collected during data preparation, to calibrating uncertainty of model prediction, to interpreting why certain behaviors of a model emerge during deployment. Specifically, ModelPred learns a parameterized function that takes a dataset $S$ as the input and predicts the model obtained by training on $S$. Our work differs from the recent work of Datamodels [1] as we aim for predicting the trained model parameters directly instead of the trained model behaviors. We demonstrate that a neural network-based set function class is capable of learning the complex relationships between the training data and model parameters. We introduce novel global and local regularization techniques to prevent overfitting and we rigorously characterize the expressive power of neural networks (NN) in approximating the end-to-end training process. Through extensive empirical investigations, we show that ModelPred enables a variety of applications that boost the interpretability and accountability of machine learning (ML), such as data valuation, data selection, memorization quantification, and model calibration.