Paper Title
Applying Neural Networks on Credit Scoring with Small Dataset and Hand-Crafted Features

Abstract
Credit scoring is importantin financial industry because of its economic importance. A recent study shows that ensemble methods that combine multiple base learnersoutperform individual learners. However, the pipeline to harmony heterogeneous base learners is complex.This work aims to apply neural network techniques to develop anindividual credit scoring learner. The resultant model achieves comparable performance to heterogeneous ensemble methods with simpler construction process and shorter training time. In addition to the model, a contribution of this work is to demonstrate the potential as well as provide a guideline to applying neural networks on small data with hand-crafted features. Keywords - Credit risk, credit scoring, machine learning, neural network