An Empirical Evaluation of Data Augmentation Strategies for Improving Model Generalization

Anthony Ndungu

Authors

Anthony Ndungu Kenya Author

Keywords:

Data augmentation, model generalization, machine learning, empirical evaluation, mixup, robustness

Abstract

In machine learning, data augmentation has emerged as a pivotal strategy to enhance model generalization, particularly when labeled data is scarce. This paper empirically evaluates various augmentation strategies, comparing their effectiveness across multiple datasets and architectures. Our study investigates augmentation techniques such as rotation, flipping, color jittering, and mixup, focusing on their impact on accuracy, robustness, and training efficiency. Results show that augmentation strategies significantly enhance model performance, with mixup achieving the highest accuracy improvement of 8.5% on average. This work contributes to understanding how augmentation can be effectively leveraged for diverse machine learning tasks.

References

Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "ImageNet Classification with Deep Convolutional Neural Networks." Advances in Neural Information Processing Systems, vol. 25, 2012, pp. 1097–1105.

Zhang, Hongyi, et al. "Mixup: Beyond Empirical Risk Minimization." ArXiv Preprint, arXiv:1710.09412, 2018.

Lekkala, C. (2020). Leveraging Lambda Architecture for Efficient Real-Time Big Data Analytics. European Journal of Advances in Engineering and Technology, 7(2), 59–64.

Goodfellow, Ian, et al. "Generative Adversarial Networks." Advances in Neural Information Processing Systems, vol. 27, 2014, pp. 2672–2680.

Yun, Sangdoo, et al. "CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features." Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019.

Lekkala, C. (2020). Advancements in Data Ingestion: Building High-Throughput Pipelines with Kafka and Spark Streaming. Journal of Scientific and Engineering Research, 7(7), 253–259.

He, Kaiming, et al. "Deep Residual Learning for Image Recognition." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2016.

Simonyan, Karen, and Andrew Zisserman. "Very Deep Convolutional Networks for Large-Scale Image Recognition." Proceedings of the International Conference on Learning Representations (ICLR), 2015.

Lekkala, C. (2020). Building Resilient Big Data Pipelines with Delta Lake for Improved Data Governance. European Journal of Advances in Engineering and Technology, 7(12), 101–106.

Shorten, Connor, and Taghi M. Khoshgoftaar. "A Survey on Image Data Augmentation for Deep Learning." Journal of Big Data, vol. 6, no. 1, 2019, pp. 1–48.

Cubuk, Ekin D., et al. "AutoAugment: Learning Augmentation Strategies from Data." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 113–123.

Lekkala, C. (2021). Modernizing Legacy Data Infrastructure for Financial Services. International Journal of Science and Research (IJSR), 10(1), 1634–1638.

Perez, Luis, and Jason Wang. "The Effectiveness of Data Augmentation in Image Classification Using Deep Learning." ArXiv Preprint, arXiv:1712.04621, 2017.

Devries, Terrance, and Graham W. Taylor. "Improved Regularization of Convolutional Neural Networks with Cutout." ArXiv Preprint, arXiv:1708.04552, 2017.

Lekkala, C. (2021). Designing High-performance, Scalable Kafka Clusters for Realtime Data Streaming. European Journal of Advances in Engineering and Technology, 8(1), 76–82.

Hernández-García, Álvaro, and Peter König. "Data Augmentation Instead of Explicit Regularization." ArXiv Preprint, arXiv:1806.03852, 2018.

Howard, Andrew G. "Some Improvements on Deep Convolutional Neural Network Based Image Classification." ArXiv Preprint, arXiv:1312.5402, 2013.