首页/论文/2007.04662

arXiv 2007.046622020-07-09cs.LGstat.ML

Untapped Potential of Data Augmentation: A Domain Generalization Viewpoint

Indian Institute of Technology, Bombay（印度理工学院孟买分校）；University of Massachusetts, Amherst（马萨诸塞大学阿默斯特分校）

Vihari Piratla, Shiv Shankar

备注：: 6 pages, ICML 2020 Workshop on Uncertainty and Ro-bustness in Deep Learning

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

英文摘要相关论文

英文摘要：

Data augmentation is a popular pre-processing trick to improve generalization accuracy. It is believed that by processing augmented inputs in tandem with the original ones, the model learns a more robust set of features which are shared between the original and augmented counterparts. However, we show that is not the case even for the best augmentation technique. In this work, we take a Domain Generalization viewpoint of augmentation based methods. This new perspective allowed for probing overfitting and delineating avenues for improvement. Our exploration with the state-of-art augmentation method provides evidence that the learned representations are not as robust even towards distortions used during training. This suggests evidence for the untapped potential of augmented examples.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

阅读原文

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

本文内容

英文摘要相关论文

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

继续探索

相关论文

VLA模型比你想象的更通用：重新审视物理和空间建模

VLA Models Are More Generalizable Than You Think: Revisiting Physical and Spatial Modeling

Sun Yat-sen University（中山大学）；Guangdong Key Laboratory of Big Data Analysis and Processing（广东省大数据分析与处理重点实验室）；X-Era AI Lab（X-Era AI实验室）

研究指出VLA模型在新视角和视觉扰动下表现下降源于空间建模不对齐，提出轻量级一 shot 调整框架，通过FTM和FLA提升模型鲁棒性，验证了少量视觉适应即可恢复视角泛化能力。

2026-03-31cs.AIcs.LGcs.RO

伪非线性数据增强：一种约束能量最小化视角

Pseudo-Nonlinear Data Augmentation: A Constrained Energy Minimization Viewpoint

University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；National Institute of Informatics（日本信息处理研究所）；SOKENDAI

本文提出了一种基于约束能量最小化视角的伪非线性数据增强方法，通过构建几何感知的潜在空间，实现了高效的数据增强与细粒度可控性。

2026-01-27cs.LG

SanitAIs: Unsupervised Data Augmentation to Sanitize Trojaned Neural Networks

Johns Hopkins University（约翰斯·霍普金斯大学）；Applied Physics Laboratory（应用物理实验室）

Self-supervised learning (SSL) methods have resulted in broad improvements to neural network performance by leveraging large, untapped collections of unlabeled data to learn generalized underlying structure. In this work, we harness unsupervised data augmentation (UDA), an SSL technique, to mitigate backdoor or Trojan attacks on deep neural networks. We show that UDA is more effective at removing trojans than current state-of-the-art methods for both feature space and point triggers, over a range of model architectures, trojans, and data quantities provided for trojan removal. These results de…

2022-06-01cs.LG

面向高效酒标识别的单图像驱动3D视角训练数据增强

Single-image driven 3d viewpoint training data augmentation for effective wine label recognition

National Yang Ming Chiao Tung University（阳明交通大学）；University of Washington（华盛顿大学）

针对酒标识别训练数据不足的问题，提出单图像驱动的3D视角数据增强方法，结合ViT与批量全三元组度量学习实现高准确率的单样本酒标识别。

2024-04-12cs.CVcs.LG

AugmentTRAJ：一个基于点的轨迹数据增强框架

AugmentTRAJ: A framework for point-based trajectory data augmentation

Memorial University of Newfoundland and Labrador（纽芬兰纪念大学）

本文提出了开源Python3框架AugmenTRAJ，通过提供可靠且保留原始数据特征的逐点数据增强方法生成合成轨迹，解决了轨迹数据增强潜力未开发的问题，提升了移动性数据分析模型的性能与泛化能力。

2023-11-25cs.LG

论混杂下的反事实数据增强

On Counterfactual Data Augmentation Under Confounding

Indian Institute of Technology Hyderabad（印度理工学院海得拉巴分校）；Microsoft Research（微软研究院）

本文从因果视角分析混杂偏差对下游分类器的影响，提出一种简单有效的反事实图像生成算法以消除混杂效应，帮助模型学习不变特征并提升泛化能力。

2023-11-21cs.CVcs.LGstat.ML

更多相关论文收起更多论文

数据增强的好、坏与丑陋一面：隐式谱正则化视角

The good, the bad and the ugly sides of data augmentation: An implicit spectral regularization perspective

Georgia Institute of Technology（佐治亚理工学院）；Samsung Research America（三星美国研究院）

本文提出理论框架，从隐式谱正则化角度解释数据增强如何通过改变协方差谱影响线性模型泛化。

2024-02-27cs.LGstat.ML

D-CODA：用于协调双臂数据增强的扩散方法

D-CODA: Diffusion for Coordinated Dual-Arm Data Augmentation

Department of Computer Science, University of Southern California（计算机科学系，南加州大学）

本文提出D-CODA，利用扩散模型为眼在手上双手模仿学习生成视角一致的双臂腕部图像及关节动作标签，并通过约束优化保证接触约束，在仿真与真实任务中验证了其优于基线方法。

2025-08-16cs.AIcs.CVcs.LG

基于零样本新视角合成的视角不变策略学习

View-Invariant Policy Learning via Zero-Shot Novel View Synthesis

Stanford University（斯坦福大学）；Toyota Research Institute（丰田研究中心）

本研究提出利用零样本新视角合成模型进行数据增强（VISTA），从单视角演示数据中学习视角不变的视觉运动策略，显著提升了策略对分布外相机视角的鲁棒性。

2025-05-31cs.AIcs.CVcs.LG

超越EAR：一种稳定的 eyelid 角度度量用于驾驶员困倦检测和数据增强

Blinking Beyond EAR: A Stable Eyelid Angle Metric for Driver Drowsiness Detection and Data Augmentation

Institue of Smart Engineering and Machine Elements at the Hamburg University of Technology（汉堡技术大学智能工程与机械元件研究所）；Intelligent Transport Systems group at the Australian Center for Robotics, The University of Sydney（澳大利亚机器人中心智能交通系统组）

本文提出了一种稳定的眼睑角度度量ELA，用于提升驾驶员困倦检测的可靠性，并通过合成数据增强扩展训练数据多样性。

2025-11-24cs.CVcs.LGeess.IV

生成性时空数据增强

Generative Spatiotemporal Data Augmentation

University of Chicago（芝加哥大学）；University of Michigan, Ann Arbor（安娜堡大学）；DEVCOM Army Research Laboratory（陆军研究实验室）

本文提出利用视频扩散模型生成时空数据以增强模型性能，尤其在数据稀缺情况下提升模型表现。

2025-12-14cs.CVcs.LG

Simple and Effective Synthesis of Indoor 3D Scenes

Google Research（谷歌研究院）；Georgia Institute of Technology（佐治亚理工学院）；University of Michigan（密歇根大学）；Apple（苹果公司）

We study the problem of synthesizing immersive 3D indoor scenes from one or more images. Our aim is to generate high-resolution images and videos from novel viewpoints, including viewpoints that extrapolate far beyond the input images while maintaining 3D consistency. Existing approaches are highly complex, with many separately trained stages and components. We propose a simple alternative: an image-to-image GAN that maps directly from reprojections of incomplete point clouds to full high-resolution RGB-D images. On the Matterport3D and RealEstate10K datasets, our approach significantly outper…

2022-12-01cs.AIcs.CVcs.LG

Data augmentation and image understanding

Institute of Cognitive Science（认知科学研究所）；University of Osnabrück（奥斯纳布吕克大学）

Interdisciplinary research is often at the core of scientific progress. This dissertation explores some advantageous synergies between machine learning, cognitive science and neuroscience. In particular, this thesis focuses on vision and images. The human visual system has been widely studied from both behavioural and neuroscientific points of view, as vision is the dominant sense of most people. In turn, machine vision has also been an active area of research, currently dominated by the use of artificial neural networks. This work focuses on learning representations that are more aligned with…

2020-12-28cs.CVcs.LG

Pose6DAug: 用于机器人数据增强的物理合理多视图物体替换

Pose6DAug: Physically Plausible Multi-view Object Swapping for Robot Data Augmentation

KAIST（韩国科学技术院）；Korea University（韩国大学）；RLWRLD

提出Pose6DAug，一种基于失败驱动的数据增强框架，通过3D网格和6D姿态轨迹替换成功轨迹中的物体，生成多视图一致的物理合理演示，无需额外数据收集，在新型物体上提升VLA策略成功率16.5%。

2026-06-23cs.LGcs.RO

Mutual Information Learned Classifiers: an Information-theoretic Viewpoint of Training Deep Learning Classification Systems

Hologic Inc（豪洛捷公司）；University of Iowa（爱荷华大学）；National University of Singapore（新加坡国立大学）；University of California at Irvine（加州大学欧文分校）；Stanford University（斯坦福大学）

Deep learning systems have been reported to acheive state-of-the-art performances in many applications, and one of the keys for achieving this is the existence of well trained classifiers on benchmark datasets which can be used as backbone feature extractors in downstream tasks. As a main-stream loss function for training deep neural network (DNN) classifiers, the cross entropy loss can easily lead us to find models which demonstrate severe overfitting behavior when no other techniques are used for alleviating it such as data augmentation. In this paper, we prove that the existing cross entrop…

2022-10-03cs.ITcs.LGmath.IT