Summary of A Survey on Data Augmentation in Large Model Era, by Yue Zhou et al.
A Survey on Data Augmentation in Large Model Era
by Yue Zhou, Chenlu Guo, Xu Wang, Yi Chang, Yuan Wu
First submitted to arxiv on: 27 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary Large language and diffusion models have demonstrated impressive capabilities in approximating human-level intelligence, attracting significant interest from academia and industry. However, training these large models requires vast amounts of high-quality data, which may soon be depleted. To address this challenge, researchers have focused on developing data augmentation methods leveraging large models. These techniques have outperformed traditional approaches, leading to a surge in research on large model-driven data augmentation methods. This paper provides an exhaustive review of these methods, categorizing relevant studies into three main categories: image augmentation, text augmentation, and paired data augmentation. The discussion also covers various post-processing techniques pertinent to large model-based data augmentation and evaluates successes and limitations across different scenarios. The authors highlight potential challenges and avenues for future exploration in the field of data augmentation, aiming to provide critical insights for researchers contributing to the advancement of more sophisticated large models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine if computers could understand us like humans do! Researchers have been working on making AI smarter by training big language and image models. However, these models need a lot of good quality data to learn from. As we keep updating these models, it’s like we’re running out of new things for them to learn. To solve this problem, scientists are finding ways to make the same amount of data go further. This paper looks at all the different ways people have been doing this and what works best. It also talks about how these techniques can be used in fields like language processing, computer vision, and audio signal processing. |
Keywords
* Artificial intelligence * Data augmentation * Diffusion * Signal processing
