Saved in:
Bibliographic Details
Main Authors: Lu, Haiquan, Liu, Xiaotian, Zhou, Yefan, Li, Qunli, Keutzer, Kurt, Mahoney, Michael W., Yan, Yujun, Yang, Huanrui, Yang, Yaoqing
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2407.12996
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916328792653824
author Lu, Haiquan
Liu, Xiaotian
Zhou, Yefan
Li, Qunli
Keutzer, Kurt
Mahoney, Michael W.
Yan, Yujun
Yang, Huanrui
Yang, Yaoqing
author_facet Lu, Haiquan
Liu, Xiaotian
Zhou, Yefan
Li, Qunli
Keutzer, Kurt
Mahoney, Michael W.
Yan, Yujun
Yang, Huanrui
Yang, Yaoqing
contents Recent studies on deep ensembles have identified the sharpness of the local minima of individual learners and the diversity of the ensemble members as key factors in improving test-time performance. Building on this, our study investigates the interplay between sharpness and diversity within deep ensembles, illustrating their crucial role in robust generalization to both in-distribution (ID) and out-of-distribution (OOD) data. We discover a trade-off between sharpness and diversity: minimizing the sharpness in the loss landscape tends to diminish the diversity of individual members within the ensemble, adversely affecting the ensemble's improvement. The trade-off is justified through our theoretical analysis and verified empirically through extensive experiments. To address the issue of reduced diversity, we introduce SharpBalance, a novel training approach that balances sharpness and diversity within ensembles. Theoretically, we show that our training strategy achieves a better sharpness-diversity trade-off. Empirically, we conducted comprehensive evaluations in various data sets (CIFAR-10, CIFAR-100, TinyImageNet) and showed that SharpBalance not only effectively improves the sharpness-diversity trade-off, but also significantly improves ensemble performance in ID and OOD scenarios.
format Preprint
id arxiv_https___arxiv_org_abs_2407_12996
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Sharpness-diversity tradeoff: improving flat ensembles with SharpBalance
Lu, Haiquan
Liu, Xiaotian
Zhou, Yefan
Li, Qunli
Keutzer, Kurt
Mahoney, Michael W.
Yan, Yujun
Yang, Huanrui
Yang, Yaoqing
Machine Learning
Recent studies on deep ensembles have identified the sharpness of the local minima of individual learners and the diversity of the ensemble members as key factors in improving test-time performance. Building on this, our study investigates the interplay between sharpness and diversity within deep ensembles, illustrating their crucial role in robust generalization to both in-distribution (ID) and out-of-distribution (OOD) data. We discover a trade-off between sharpness and diversity: minimizing the sharpness in the loss landscape tends to diminish the diversity of individual members within the ensemble, adversely affecting the ensemble's improvement. The trade-off is justified through our theoretical analysis and verified empirically through extensive experiments. To address the issue of reduced diversity, we introduce SharpBalance, a novel training approach that balances sharpness and diversity within ensembles. Theoretically, we show that our training strategy achieves a better sharpness-diversity trade-off. Empirically, we conducted comprehensive evaluations in various data sets (CIFAR-10, CIFAR-100, TinyImageNet) and showed that SharpBalance not only effectively improves the sharpness-diversity trade-off, but also significantly improves ensemble performance in ID and OOD scenarios.
title Sharpness-diversity tradeoff: improving flat ensembles with SharpBalance
topic Machine Learning
url https://arxiv.org/abs/2407.12996