Approaches to Fraud Detection - Autoencoder and Isolation Forest - Fraud Detection Using ML

Explore fraud detection techniques using machine learning in this comprehensive webinar. Learn to combat fraud using KNIME, a free low-code tool, without writing code or relying on if-then rules. Discover approaches for both labeled and unlabeled data, including random forest, autoencoders, visualizations, and statistical methods. Master the use of Isolation Forest and DBSCAN algorithms for detecting fraudulent activity. Gain practical skills in implementing various fraud detection techniques using the KNIME Analytics Platform, from basic decision trees to advanced deep learning methods.

Introduction
KNIME Analytics Platform
KNIME nodes & workflow
Goals for the Session
Fraud is all around us
Potentially fraudulent data
Fraudulent data might be labelled
Decision Tree
Random Forest
Advanced: Sampling Strategies
Finding fraud through deep learning
A neural autoencoder in KNIME
Walk through how to do the same task using unlabeled data Jinwei
Fraud and Outlier Detection
Finding Outliers: Statistics
Demo IQR and Z-score Implementation in KNIME
DBSCAN
Summary
Useful Fraud-related links
Useful KNIME-related links
Q&A