Copyright © 2025 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{165489,
        author = {Nandan Parmar},
        title = {Ensemble of Data Augmentation Techniques for Efficient 3 Augmentation in NLP},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {1},
        pages = {2706-2734},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=165489},
        abstract = {In the last decade, NLP has made 
significant advances in machine learning. In so many 
machine learning scenarios, there isn't enough data 
available to train a good classifier. Data augmentation 
can indeed be utilized to solve this problem. It utilizes 
transformations to artificially increase the amount of 
available training data. Due of linguistic data's 
discrete character, this topic is still relatively 
underexplored, in spite of the huge rise in usage. A 
major goal of the DA techniques is to increase the 
diversity of training data, allowing the model to 
better generalize when faced with novel testing data. 
This study uses the term "data augmentation" to 
allude as a broad concept that encompasses 
techniques for transforming training data. While 
most text data augmentation research focuses on the 
long-term aim of developing end-to-end learning 
solutions, this study focuses on using pragmatic, 
robust, 
scalable, 
and easy-to-implement data 
augmentation techniques comparable to those used in 
computer vision. In natural language processing, 
simple but successful data augmentation procedures 
have been implemented and inspired by such efforts, 
we construct and compare ensemble data 
augmentation for NLP classification. We are 
proposing an ensembling of simple yet effective data 
augmentation techniques. Through experiments on 
various dataset from kaggle, we show that ensembling 
of augmentation can boost performance with any text 
embedding technique particularly for small training 
sets. We conclude by carrying out experiments on a 
classification datasets. Based on the results, we draw 
conclusion that Effective DA approach by ensembles 
of data augmentation can help practitioners choose 
suitable augmentation technique in different settings.},
        keywords = {Text Data Augmentation, NLP, Class  Imbalance, Text Embeddings},
        month = {June},
        }