Data Balancing with Gen AI: Credit Card Fraud Detection

Overview

Class Central Tips

In this 2-hour guided project, you will learn how to leverage Generative AI for data generation to address data imbalance. SecureTrust Financial Services, a financial institution, has asked us to help them improve the accuracy of their fraud detection system. The model is a binary classifier, but it's not performing well due to data imbalance. As data scientists, we will employ Generative Adversarial Networks (GANs), a subset of Generative AI, to create synthetic fraudulent transactions that closely resemble real transactions. This approach aims to balance the dataset and enhance the accuracy of the fraud detection model. This guided project is designed for those interested in learning how Generative models can increase model accuracy by generating synthetic data. To make the most of this project, it is recommended to have at least one year of experience using deep learning frameworks such as TensorFlow and Keras in Python.

Syllabus

Project Overview

A financial institution, called SecureTrust Financial Services, has contracted us to improve the accuracy of their fraud detection machine learning model. The model is a binary classifier, but it is not working well because the data is imbalanced. To solve this problem as data scientists, we will use generative adversarial networks (GANs), a type of Generative AI, to generate synthetic fraudulent transactions that are indistinguishable from real transactions. This will help to balance the dataset and improve the accuracy of the fraud detection model.