Analysis of Blind Validation for Learning Models

Abhijeet Jadhav, Poonam M Shettar, Divya Karoshi, Aishwarya G, Akash Kulkarni, Basawaraj Patil

2025

Abstract

Machine learning, deep learning, and image processing have gained significant traction and are widely utilized across various fields for many applications, contributing significantly to advancements in medicine, security, robotics, automation, and beyond. With the expanding range of applications, it is crucial to comprehend how models perform on unseen data and assess their reliability for deployment in real-time scenarios. In this context, this study employs CNN models to provide insights on blind validation. Convolutional Neural Network (CNN) models have emerged as powerful tools for image classification tasks. This study employs a CNN model for digit classification using the famous MNIST dataset. Subsequently, the model’s performance is evaluated on two additional datasets: USPS and EMNIST. The evaluation aims to understand how the model generalizes across different datasets with varying characteristics and to assess its robustness in real-world applications. Blind validation is conducted by training the model on the MNIST dataset and testing it on itself and the other datasets to observe potential biases and inconsistencies in the model’s behaviour across diverse datasets. This analysis provides valuable insights into the model’s adaptability and reliability for deployment in practical scenarios beyond the training dataset’s domain.

Paper Citation

in Harvard Style

Jadhav A., M Shettar P., Karoshi D., G A., Kulkarni A. and Patil B. (2025). Analysis of Blind Validation for Learning Models. In Proceedings of the 3rd International Conference on Futuristic Technology - Volume 3: INCOFT; ISBN 978-989-758-763-4, SciTePress, pages 295-300. DOI: 10.5220/0013614300004664

in Bibtex Style

@conference{incoft25,
author={Abhijeet Jadhav and Poonam M Shettar and Divya Karoshi and Aishwarya G and Akash Kulkarni and Basawaraj Patil},
title={Analysis of Blind Validation for Learning Models},
booktitle={Proceedings of the 3rd International Conference on Futuristic Technology - Volume 3: INCOFT},
year={2025},
pages={295-300},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013614300004664},
isbn={978-989-758-763-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 3rd International Conference on Futuristic Technology - Volume 3: INCOFT
TI - Analysis of Blind Validation for Learning Models
SN - 978-989-758-763-4
AU - Jadhav A.
AU - M Shettar P.
AU - Karoshi D.
AU - G A.
AU - Kulkarni A.
AU - Patil B.
PY - 2025
SP - 295
EP - 300
DO - 10.5220/0013614300004664
PB - SciTePress