This presentation discusses how Machine Learning can be applied to Payments to respond rapidly to known and emerging patterns of fraud, and to detect patterns of fraud that may not otherwise be identified.
It will cover techniques that have been used and are emerging in fraud detection including rule-based techniques, supervised learning and unsupervised learning. The presentation includes a demonstration using TensorFlow to detect fraud. This will illustrate the process of preparing training and test data, learning and then applying the model to generate potential fraud events.
We’ll also explore potential issues including data bias and mitigating approaches.
Develop an understanding of how machine learning can be implemented to detect fraud in payments, gain knowledge on how Machine Learning can detect patterns that indicate fraudulent transactions, learn how data is prepared to train an artificial Neural Network
Basic familiarity with machine learning is helpful, but not required
You can see Tamsin’s slides below: