Credit Approval System using Semi-Supervised Learning

A variety of supervised and unsupervised learning
algorithms from the fields of machine learning and pattern
recognition have been employed to enhance the efficiency of credit
approval systems. Supervised learning relies solely on labeled
data for training classifiers, which can be a time-consuming
process that often requires input from domain experts. In
contrast, unlabeled data is typically more accessible in many
real-world scenarios. Semi-supervised learning (SSL) effectively
addresses this challenge by combining a limited amount of labeled
data with a larger set of unlabeled examples. This dual approach
allows SSL to create stronger and more accurate classifiers than
traditional supervised learning methods alone. By utilizing both
data types, SSL capitalizes on the information contained within
unlabeled samples, thereby enhancing the model’s performance
and generalization ability. In this research, we have utilized self-
training and fuzziness-based semi-supervised learning strategies
tailored for credit approval systems within the banking and
finance sectors. Our experiments demonstrate that leveraging
samples categorized by self-training and fuzzy SSL significantly
enhances the overall accuracy and robustness of the credit
approval process compared to supervised classifiers.