Data Augmentation
Data augmentation is a technique used to artificially expand datasets by creating modified versions of existing data, primarily to improve the performance and robustness of machine learning models, especially when training data is scarce. Current research focuses on developing more sophisticated augmentation methods, including those leveraging generative models like GANs and diffusion models, and integrating augmentation with other techniques such as contrastive learning and transfer learning, often applied within architectures like transformers and convolutional neural networks. This work is significant because it addresses the limitations of limited datasets across various domains, from image classification and object detection to natural language processing and time series forecasting, leading to more accurate and generalizable models for diverse applications.
Papers
Reinforcement Learning with Euclidean Data Augmentation for State-Based Continuous Control
Jinzhu Luo, Dingyang Chen, Qi Zhang
Long-Tailed Backdoor Attack Using Dynamic Data Augmentation Operations
Lu Pang, Tao Sun, Weimin Lyu, Haibin Ling, Chao Chen
Explanation-Preserving Augmentation for Semi-Supervised Graph Representation Learning
Zhuomin Chen, Jingchao Ni, Hojat Allah Salehi, Xu Zheng, Esteban Schafir, Farhad Shirani, Dongsheng Luo
Feature Augmentation for Self-supervised Contrastive Learning: A Closer Look
Yong Zhang, Rui Zhu, Shifeng Zhang, Xu Zhou, Shifeng Chen, Xiaofan Chen
REFINE on Scarce Data: Retrieval Enhancement through Fine-Tuning via Model Fusion of Embedding Models
Ambuje Gupta, Mrinal Rawat, Andreas Stolcke, Roberto Pieraccini
Graph Masked Autoencoder for Spatio-Temporal Graph Learning
Qianru Zhang, Haixin Wang, Siu-Ming Yiu, Hongzhi Yin
Generative Deep Learning and Signal Processing for Data Augmentation of Cardiac Auscultation Signals: Improving Model Robustness Using Synthetic Audio
Leigh Abbott, Milan Marocchi, Matthew Fynn, Yue Rong, Sven Nordholm