Credit and Loan Approval Classification Using a Bio-Inspired Neural Network.

Area of Science:

• Artificial Intelligence
• Machine Learning
• Computational Finance

Background:

• The banking industry faces challenges in loan and credit approval due to increasing applications and limited assets.
• Accurate risk assessment is crucial for efficient resource allocation and minimizing financial losses.
• Traditional methods struggle with the complexity and scale of modern credit scoring.

Purpose of the Study:

• To develop an advanced machine learning model for efficient and accurate credit and loan approval.
• To address the limitations of conventional neural networks, such as slow training and local minima.
• To introduce a novel bio-inspired algorithm that enhances the learning process for binary classification problems in finance.

Main Methods:

• Development of a novel weights and structure determination (WASD) neural network.
• Creation of a bio-inspired WASD algorithm for binary classification problems (BWASD).
• Integration of the metaheuristic beetle antennae search (BAS) algorithm to optimize the WASD learning procedure.

Main Results:

• The BWASD algorithm demonstrates superior performance and adaptability compared to conventional back-propagation neural networks.
• The proposed model effectively handles the unique characteristics of credit and loan approval tasks.
• Theoretical and experimental studies confirm the enhanced efficiency and reduced risk in applicant selection.

Conclusions:

• The BWASD algorithm offers a significant advancement in machine learning for financial risk assessment.
• This bio-inspired approach provides a more robust and efficient solution for credit and loan approval systems.
• A comprehensive MATLAB package is available to support the implementation and further research.