On the Generalization Effects of Linear Transformations in Data
Augmentation
- URL: http://arxiv.org/abs/2005.00695v3
- Date: Wed, 26 Jul 2023 22:58:34 GMT
- Title: On the Generalization Effects of Linear Transformations in Data
Augmentation
- Authors: Sen Wu, Hongyang R. Zhang, Gregory Valiant, Christopher R\'e
- Abstract summary: Data augmentation is a powerful technique to improve performance in applications such as image and text classification tasks.
We study a family of linear transformations and study their effects on the ridge estimator in an over-parametrized linear regression setting.
We propose an augmentation scheme that searches over the space of transformations by how uncertain the model is about the transformed data.
- Score: 32.01435459892255
- License: http://creativecommons.org/licenses/by/4.0/
- Abstract: Data augmentation is a powerful technique to improve performance in
applications such as image and text classification tasks. Yet, there is little
rigorous understanding of why and how various augmentations work. In this work,
we consider a family of linear transformations and study their effects on the
ridge estimator in an over-parametrized linear regression setting. First, we
show that transformations that preserve the labels of the data can improve
estimation by enlarging the span of the training data. Second, we show that
transformations that mix data can improve estimation by playing a
regularization effect. Finally, we validate our theoretical insights on MNIST.
Based on the insights, we propose an augmentation scheme that searches over the
space of transformations by how uncertain the model is about the transformed
data. We validate our proposed scheme on image and text datasets. For example,
our method outperforms random sampling methods by 1.24% on CIFAR-100 using
Wide-ResNet-28-10. Furthermore, we achieve comparable accuracy to the SoTA
Adversarial AutoAugment on CIFAR-10, CIFAR-100, SVHN, and ImageNet datasets.
Related papers
- Automatic Data Augmentation via Invariance-Constrained Learning [94.27081585149836]
Underlying data structures are often exploited to improve the solution of learning tasks.
Data augmentation induces these symmetries during training by applying multiple transformations to the input data.
This work tackles these issues by automatically adapting the data augmentation while solving the learning task.
arXiv Detail & Related papers (2022-09-29T18:11:01Z) - Optimizing transformations for contrastive learning in a differentiable
framework [4.828899860513713]
We propose a framework to find optimal transformations for contrastive learning using a differentiable transformation network.
Our method increases performances at low annotated data regime both in supervision accuracy and in convergence speed.
Experiments were performed on 34000 2D slices of brain Magnetic Resonance Images and 11200 chest X-ray images.
arXiv Detail & Related papers (2022-07-27T08:47:57Z) - Invariance Learning in Deep Neural Networks with Differentiable Laplace
Approximations [76.82124752950148]
We develop a convenient gradient-based method for selecting the data augmentation.
We use a differentiable Kronecker-factored Laplace approximation to the marginal likelihood as our objective.
arXiv Detail & Related papers (2022-02-22T02:51:11Z) - Self-Supervised Pre-Training for Transformer-Based Person
Re-Identification [54.55281692768765]
Transformer-based supervised pre-training achieves great performance in person re-identification (ReID)
Due to the domain gap between ImageNet and ReID datasets, it usually needs a larger pre-training dataset to boost the performance.
This work aims to mitigate the gap between the pre-training and ReID datasets from the perspective of data and model structure.
arXiv Detail & Related papers (2021-11-23T18:59:08Z) - Improving Model Generalization by Agreement of Learned Representations
from Data Augmentation [19.286766429954174]
In computer vision, data augmentation techniques such as CutOut, MixUp, and CutMix demonstrated state-of-the-art (SOTA) results.
We call our proposed method Agreement Maximization or simply AgMax.
We show that AgMax can take advantage of the data augmentation to consistently improve model generalization by a significant margin.
arXiv Detail & Related papers (2021-10-20T12:44:52Z) - Semantic Perturbations with Normalizing Flows for Improved
Generalization [62.998818375912506]
We show that perturbations in the latent space can be used to define fully unsupervised data augmentations.
We find that our latent adversarial perturbations adaptive to the classifier throughout its training are most effective.
arXiv Detail & Related papers (2021-08-18T03:20:00Z) - Learning Representational Invariances for Data-Efficient Action
Recognition [52.23716087656834]
We show that our data augmentation strategy leads to promising performance on the Kinetics-100, UCF-101, and HMDB-51 datasets.
We also validate our data augmentation strategy in the fully supervised setting and demonstrate improved performance.
arXiv Detail & Related papers (2021-03-30T17:59:49Z) - Adaptive Weighting Scheme for Automatic Time-Series Data Augmentation [79.47771259100674]
We present two sample-adaptive automatic weighting schemes for data augmentation.
We validate our proposed methods on a large, noisy financial dataset and on time-series datasets from the UCR archive.
On the financial dataset, we show that the methods in combination with a trading strategy lead to improvements in annualized returns of over 50$%$, and on the time-series data we outperform state-of-the-art models on over half of the datasets, and achieve similar performance in accuracy on the others.
arXiv Detail & Related papers (2021-02-16T17:50:51Z) - Augmentation Inside the Network [1.5260179407438161]
We present augmentation inside the network, a method that simulates data augmentation techniques for computer vision problems.
We validate our method on the ImageNet-2012 and CIFAR-100 datasets for image classification.
arXiv Detail & Related papers (2020-12-19T20:07:03Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.