MetaAugment: Sample-Aware Data Augmentation Policy Learning

Zhou, Fengwei; Li, Jiawei; Xie, Chuanlong; Chen, Fei; Hong, Liang; Sun, Rui; Li, Zhenguo

doi:10.1609/aaai.v35i12.17324

Cited by 11 publications

(10 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The instance-level approach builds a policy network that predicts an image-specific augmentation policy suitable for each input image. Zhou et al [23] used a policy network to assign different weights to transformed images in CNN training. Cheung and Yeung [24] used a policy network that adaptively generated the hyperparameters for the transformation operations of each image.…”

Section: B Optimization Of Data Augmentation Policymentioning

confidence: 99%

Class-Adaptive Data Augmentation for Image Classification

Yoo

Kang

2023

IEEE Access

View full text Add to dashboard Cite

Data augmentation is a widely used regularization technique for improving the performance of convolutional neural networks (CNNs) in image classification tasks. To improve the effectiveness of data augmentation, it is important to find label-preserving transformations that fit the domain knowledge for a given dataset. In several real-world datasets, appropriate augmentation policies differ between classes, owing to their different characteristics. In this paper, we propose a class-adaptive data augmentation method that utilizes class-specific augmentation policies. First, we train the CNN without data augmentation. Subsequently, we derive a suitable augmentation policy for each class through an optimization procedure to maximize the degree of transformation while maintaining the label-preserving property of CNNs. Finally, we re-train the model using data augmentation based on derived class-specific augmentation policies. Through experiments using benchmark datasets with class-specific transformation constraints, we demonstrate that the proposed method achieves comparable or higher classification accuracy than the baseline methods using the same augmentation policy for all classes. Additionally, we confirm that the derived class-specific augmentation policies are consistent with the domain knowledge of each dataset. INDEX TERMS image classification, data augmentation, class-adaptive data augmentation, hyperparameter optimizationRecently, attempts have been made to automatically search for an appropriate augmentation policy for a given training dataset in a data-driven manner [10][11][12][13][14][15]. Existing methods can be used to effectively apply data augmentation to improve the performance of a CNN in the absence of domain knowledge. These methods mainly focus on optimizing the augmentation policy to be dataset-specific, implying that every image in the dataset is randomly transformed in the same manner, regardless of the class label.Our research motivation stems from the fact that appropriate augmentation policies can differ between classes. For example, in the digit images of MNIST and SVHN datasets, random horizontal and vertical flips preserve the class labels

show abstract

Section: B Optimization Of Data Augmentation Policymentioning

confidence: 99%

Class-Adaptive Data Augmentation for Image Classification

Yoo

Kang

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Given an image recognition task with a training dataset D tr = {(x i , y i } |D tr | i=1 , with x i and y i representing the image and label respectively, augmented samples T (x i ) are derived by applying augmentation policy T to sample x i . Usually, the policy T is composed of multiple sub-policies τ , and each sub-policy is made up by K augmentation operations O, optionally with their corresponding probabilities and magnitudes, which are adopted in the original design of AutoAugment [1], but not included in some of the recent methods such as Weight-sharing AutoAugment [18] and MetaAugment [26].…”

Section: Conventional Augmentation Searchmentioning

confidence: 99%

“…While most previous studies focus on learning augmentation policies for the entire dataset, MetaAugment [26] proposes to learn sample-aware augmentation policies during model training by formulating the policy search as a sample reweighting problem, and constructing a policy network to learn the weights of specific augmented images by minimizing the validation loss via meta learning. Despite its benefits, MetaAugment is computationally expensive, requiring three forward and backward passes of the target network in each iteration.…”

Section: Related Workmentioning

confidence: 99%

“…For a fair comparison, we list results of stationary policies produced by static strategies, AutoAugment [1], FastAA [16], and DADA [15]. We also include results from dynamic strategies, PBA [9], AdvAA [25], and MetaAug [26], producing non-stationary policies as target model training progresses.…”

Section: Datasets Metrics and Baselinesmentioning

confidence: 99%

“…This observation implies the limitation of label or sampleinvariant dataset-level augmentation policies. MetaAugment [26] proposes to learn a sample-aware augmentation policy by solving a sample re-weighting problem. It uses an augmentation policy network to take an augmentation operation and the corresponding augmented image as inputs, and outputs a weight to adjust the augmented image loss computed by the task network.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

LA3: Efficient Label-Aware AutoAugment

Zhao

Wang

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Automated augmentation is an emerging and effective technique to search for data augmentation policies to improve generalizability of deep neural network training. Most existing work focuses on constructing a unified policy applicable to all data samples in a given dataset, without considering sample or class variations. In this paper, we propose a novel two-stage data augmentation algorithm, named Label-Aware Au-toAugment (LA3), which takes advantage of the label information, and learns augmentation policies separately for samples of different labels. LA3 consists of two learning stages, where in the first stage, individual augmentation methods are evaluated and ranked for each label via Bayesian Optimization aided by a neural predictor, which allows us to identify effective augmentation techniques for each label under a low search cost. And in the second stage, a composite augmentation policy is constructed out of a selection of effective as well as complementary augmentations, which produces significant performance boost and can be easily deployed in typical model training. Extensive experiments demonstrate that LA3 achieves excellent performance matching or surpassing existing methods on CIFAR-10 and CIFAR-100, and achieves a new state-of-the-art ImageNet accuracy of 79.97% on ResNet-50 among autoaugmentation methods, while maintaining a low computational cost.

show abstract

ORAD: a new framework of offline Reinforcement Learning with Q-value regularization

et al. 2022

View full text Add to dashboard Cite

This paper presents advanced techniques of training diffusion policies for offline reinforcement learning (RL). At the core is a mean-reverting stochastic differential equation (SDE) that transfers a complex action distribution into a standard Gaussian and then samples actions conditioned on the environment state with a corresponding reverse-time SDE, like a typical diffusion policy. We show that such an SDE has a solution that we can use to calculate the log probability of the policy, yielding an entropy regularizer that improves the exploration of offline datasets. To mitigate the impact of inaccurate value functions from out-of-distribution data points, we further propose to learn the lower confidence bound of Q-ensembles for more robust policy improvement. By combining the entropy-regularized diffusion policy with Q-ensembles in offline RL, our method achieves state-of-the-art performance on most tasks in D4RL benchmarks. Code is available at https://github.com/ruoqizzz/Entropy-Regularized-Diffusion-Policy-with-QEnsemble.

show abstract

MetaAugment: Sample-Aware Data Augmentation Policy Learning

Cited by 11 publications

References 23 publications

Class-Adaptive Data Augmentation for Image Classification

Class-Adaptive Data Augmentation for Image Classification

LA3: Efficient Label-Aware AutoAugment

ORAD: a new framework of offline Reinforcement Learning with Q-value regularization

Contact Info

Product

Resources

About