Using mixup as a regularizer can surprisingly improve accuracy and out-of-distribution robustness

Using mixup as a regularizer can surprisingly improve accuracy and out-of-distribution robustness

We show that the effectiveness of the well celebrated Mixup can be further improved if instead of using it as the sole learning objective, it is utilized as an additional regularizer to the standard cross-entropy loss. This simple change not only improves accuracy but also significantly improves the...

書誌詳細
主要な著者:	Pinto, F, Yang, H, Lim, SN, Torr, PHS, Dokania, PK
フォーマット:	Conference item
言語:	English
出版事項:	Curran Associates, Inc 2023

類似資料

Mix-MaxEnt: improving accuracy and uncertainty estimates of deterministic neural networks
著者:: Pinto, F, 等
出版事項: (2021)

Placing objects in context via inpainting for out-of-distribution segmentation
著者:: De Jorge, P, 等
出版事項: (2024)

An impartial take to the CNN vs transformer robustness contest
著者:: Pinto, F, 等
出版事項: (2022)

Raising the bar on the evaluation of out-of-distribution detection
著者:: Mukhoti, J, 等
出版事項: (2023)

Are vision transformers always more robust than convolutional neural networks?
著者:: Pinto, F, 等
出版事項: (2021)

Sample-dependent adaptive temperature scaling for improved calibration
著者:: Joy, T, 等
出版事項: (2023)

Deep facial phenotyping with mixup augmentation
著者:: Campbell, J, 等
出版事項: (2023)

Data augmentation and mixup techniques for stock returns prediction
著者:: Tan, Regan Yik Rong
出版事項: (2024)

Computation-efficient knowledge distillation via uncertainty-aware mixup
著者:: Xu, Guodong, 等
出版事項: (2023)

GDumb: A simple approach that questions our progress in continual learning
著者:: Prabhu, A, 等
出版事項: (2020)

Discovering class-specific pixels for weakly-supervised semantic segmentation
著者:: Chaudhry, A, 等
出版事項: (2017)

Semantic-discriminative mixup for generalizable sensor-based cross-domain activity recognition
著者:: Lu, Wang, 等
出版事項: (2023)

A surprisingly robust trick for the Winograd Schema Challenge
著者:: Kocijan, V, 等
出版事項: (2019)

Continual learning in low-rank orthogonal subspaces
著者:: Chaudhry, A, 等
出版事項: (2020)

Surprise
著者:: Ross, Wendy, 等
出版事項: (2022)

Surprises.
著者:: Weiskrantz, L
出版事項: (2008)

No Surprises
著者:: Wells, Ian
出版事項: (2021)

Mirror Descent view for Neural Network quantization
著者:: Ajanthan, T, 等
出版事項: (2021)

How robust is unsupervised representation learning to distribution shift?
著者:: Shi, Y, 等
出版事項: (2023)

Regularization Through Feature Knock Out
著者:: Wolf, Lior, 等
出版事項: (2005)

On surprise-hacking
著者:: Felin, T, 等
出版事項: (2019)

Integral sliding mode control for improved robustness and accuracy of induction motors
著者:: Mohamed, H.A.F., 等
出版事項: (2008)

AirMixML: Over-the-Air Data Mixup for Inherently Privacy-Preserving Edge Machine Learning
著者:: Koda, Yusuke, 等
出版事項: (2022)

The Surprising Breadth of Harbingers of Failure
著者:: Simester, Duncan, 等
出版事項: (2021)

Robust parameterization and computation of the trifocal tensor
著者:: Torr, PHS, 等
出版事項: (1998)

STEER: simple temporal regularization for neural ODEs
著者:: Ghosh, A, 等
出版事項: (2020)

On using focal loss for neural network calibration
著者:: Mukhoti, J, 等
出版事項: (2020)

Calibrating deep neural networks using focal loss
著者:: Mukhoti, J, 等
出版事項: (2020)

RanDumb: a simple approach that questions the efficacy of continual representation learning
著者:: Prabhu, A, 等
出版事項: (2025)

RanDumb: random representations outperform online continually learned representations
著者:: Prabhu, A, 等
出版事項: (2025)

Supervised contrastive pretrained ResNet with MixUp to enhance respiratory sound classification on imbalanced and limited dataset
著者:: Hu, Jinhai, 等
出版事項: (2024)

Surprise, surprise! (again) The 2017 British general election exit poll
著者:: Curtice, J, 等
出版事項: (2017)

Jackie surprises himself
著者:: Singh, Devinder
出版事項: (2016)

The surprise theory of everything
著者:: Vedral, V
出版事項: (2012)

Aristophanes and the poetics of surprise
著者:: Kanellakis, D
出版事項: (2019)

Surprises in theoretical physics/
著者:: 406046 Peierls, Rudolf
出版事項: (1979)

Surprises in the Consumption Function.
著者:: Muellbauer, J
出版事項: (1983)

Universal regularization methods: varying the power, the smoothness and the accuracy
著者:: Cartis, C, 等
出版事項: (2019)

Regularity and robustness in monotone Bayesian games
著者:: Beggs, A
出版事項: (2015)

Regularity and robustness in monotone Bayesian games
著者:: Beggs, A
出版事項: (2015)