# On the Reproducibility of Neural Network Predictions

Source: [On the Reproducibility of Neural Network Predictions](<https://www.sota2.com/research/paper/68fa693c1e209131cea210c4>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-25T22:56:51.523Z

Authors: Srinadh Bhojanapalli, Kimberly Wilber, Andreas Veit, Ankit Singh Rawat, Seungyeon Kim, Aditya Menon, Sanjiv Kumar

Published: 2021-02-05T18:51:01.000Z

## Abstract

Standard training techniques for neural networks involve multiple sources of randomness, e.g., initialization, mini-batch ordering and in some cases data augmentation. Given that neural networks are heavily over-parameterized in practice, such randomness can cause {\\em churn} -- for the same input, disagreements between predictions of the two models independently trained by the same algorithm, contributing to the \`reproducibility challenges' in modern machine learning. In this paper, we study this problem of churn, identify factors that cause it, and propose two simple means of mitigating it. We first demonstrate that churn is indeed an issue, even for standard image classification tasks (CIFAR and ImageNet), and study the role of the different sources of training randomness that cause churn. By analyzing the relationship between churn and prediction confidences, we pursue an approach with two components for churn reduction. First, we propose using \\emph{minimum entropy regularizers} to increase prediction confidences. Second, \\changes{we present a novel variant of co-distillation approach~\\citep{anil2018large} to increase model agreement and reduce churn}. We present empirical results showing the effectiveness of both techniques in reducing churn while improving the accuracy of the underlying model.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2102.03349>)

[PDF](<https://arxiv.org/pdf/2102.03349.pdf>)
