Summary of Reducing and Exploiting Data Augmentation Noise Through Meta Reweighting Contrastive Learning For Text Classification, by Guanyi Mou et al.
Reducing and Exploiting Data Augmentation Noise through Meta Reweighting Contrastive Learning for Text Classification
by Guanyi Mou, Yichuan Li, Kyumin Lee
First submitted to arxiv on: 26 Sep 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG)
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 The proposed novel framework leverages meta learning and contrastive learning techniques to reweight augmented samples in text classification tasks, refining their feature representations based on quality. The framework incorporates novel weight-dependent enqueue and dequeue algorithms, demonstrating the ability to cooperate with existing deep learning models like RoBERTa-base and Text-CNN, as well as augmentation techniques such as Wordnet and Easydata. Experimental results show significant improvements of up to 4.3% absolute improvement on seven GLUE benchmark datasets compared to the best baseline. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper proposes a new way to improve deep learning models’ performance when using augmented data in text classification tasks. The idea is to use special algorithms that look at how good or bad each piece of augmented data is, and adjust the model’s behavior based on this information. This helps the model learn more effectively from the high-quality data and ignore the low-quality data. The results show that this approach can lead to significant improvements in performance, with an average improvement of 1.6-4.3% compared to the best previous methods. |
Keywords
» Artificial intelligence » Cnn » Deep learning » Meta learning » Text classification
