TY  - JOUR
AU  - Myakala, Praveen Kumar
PY  - 2025
DA  - 2025/03/15
TI  - Scaling AI with Limited Labeled Data: A Self-Supervised Learning Approach
JO  - ICCK Transactions on Emerging Topics in Artificial Intelligence
T2  - ICCK Transactions on Emerging Topics in Artificial Intelligence
JF  - ICCK Transactions on Emerging Topics in Artificial Intelligence
VL  - 2
IS  - 1
SP  - 26
EP  - 35
DO  - 10.62762/TETAI.2025.607708
UR  - https://www.icck.org/article/abs/TETAI.2025.607708
KW  - self-supervised Learning (SSL)
KW  - limited labeled data
KW  - data-scarce scenarios
KW  - contrastive learning
KW  - masked autoencoding
KW  - scalable AI
AB  - The scalability of modern AI is fundamentally limited by the availability of labeled data. While supervised learning achieves remarkable performance, it relies on large annotated datasets, which are expensive and time-consuming to acquire. This work explores self-supervised learning (SSL) as a promising solution to this challenge, enabling AI to scale effectively in data-scarce scenarios. This study demonstrates the effectiveness of the proposed SSL framework using the EuroSAT dataset, a benchmark for land cover classification where labeled data is limited and costly. The proposed approach integrates contrastive learning with multi-spectral augmentations, such as spectral jittering and band shuffling, along with masked autoencoding that applies spatial-spectral masking based on local variance in spectral bands. This method effectively captures the unique spatial and spectral characteristics of EuroSAT imagery. Experimental results show that the proposed SSL-based models achieve 81.2% accuracy with only 10% of the labeled data, outperforming supervised learning by 2.7% and semi-supervised methods by 2.1%. These results demonstrate the potential of SSL to reduce reliance on labeled data and enable effective AI deployment in data-constrained environments. The proposed work highlights the transformative potential of SSL in reducing annotation burdens, paving the way for more scalable, accessible, and cost-effective AI solutions.
SN  - 3068-6652
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Myakala2025Scaling,
  author = {Praveen Kumar Myakala},
  title = {Scaling AI with Limited Labeled Data: A Self-Supervised Learning Approach},
  journal = {ICCK Transactions on Emerging Topics in Artificial Intelligence},
  year = {2025},
  volume = {2},
  number = {1},
  pages = {26-35},
  doi = {10.62762/TETAI.2025.607708},
  url = {https://www.icck.org/article/abs/TETAI.2025.607708},
  abstract = {The scalability of modern AI is fundamentally limited by the availability of labeled data. While supervised learning achieves remarkable performance, it relies on large annotated datasets, which are expensive and time-consuming to acquire. This work explores self-supervised learning (SSL) as a promising solution to this challenge, enabling AI to scale effectively in data-scarce scenarios. This study demonstrates the effectiveness of the proposed SSL framework using the EuroSAT dataset, a benchmark for land cover classification where labeled data is limited and costly. The proposed approach integrates contrastive learning with multi-spectral augmentations, such as spectral jittering and band shuffling, along with masked autoencoding that applies spatial-spectral masking based on local variance in spectral bands. This method effectively captures the unique spatial and spectral characteristics of EuroSAT imagery. Experimental results show that the proposed SSL-based models achieve 81.2\% accuracy with only 10\% of the labeled data, outperforming supervised learning by 2.7\% and semi-supervised methods by 2.1\%. These results demonstrate the potential of SSL to reduce reliance on labeled data and enable effective AI deployment in data-constrained environments. The proposed work highlights the transformative potential of SSL in reducing annotation burdens, paving the way for more scalable, accessible, and cost-effective AI solutions.},
  keywords = {self-supervised Learning (SSL), limited labeled data, data-scarce scenarios, contrastive learning, masked autoencoding, scalable AI},
  issn = {3068-6652},
  publisher = {Institute of Central Computation and Knowledge}
}