SimCLR vs. Supervised ResNet for Low-Label Image Classification with Data Augmentation

Varghese, Jeson Varghese (2025) SimCLR vs. Supervised ResNet for Low-Label Image Classification with Data Augmentation. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (931kB) | Preview

Abstract

The success of deep learning in computer vision is often hindered by the scarcity of large, annotated datasets. This study empirically evaluates the SimCLR self-supervised learning framework against a supervised ResNet on CIFAR-10 under label-scarce conditions (1augmentation, a key component of contrastive learning. Findings reveal that robust data augmentation is not merely beneficial but essential for SimCLR’s success. With strong augmentations, SimCLR significantly outperforms the supervised baseline in all low-label settings, establishing its efficacy for learning powerful representations when labeled data is limited.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Hamill, David UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence > Computer vision Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence > Computer vision Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	24 Aug 2026 12:33
Last Modified:	24 Aug 2026 12:33
URI:	https://norma.ncirl.ie/id/eprint/9607

Actions (login required)

View Item