A graph neural network approach for time evolving fraud networks
Who is this presentation for?Data scientists or analysts
Adobe has an e-commerce platform where it sells various subscription products like Creative Cloud, Photoshop and Stock. We encounter large volume of fraudulent activity such as card testing, trial abuse, seat addition among others. In this talk we will describe how we represent the data as a heterogenous graph and use state-of-art graph neural network based approach to identify fraud.
We will start by introducing various types of fraud in eCommerce. Then, we will introduce our data, talk about its characteristics and how we represent it as a heterogenous attributed graph. Next, we will introduce the audience to state-of-the-art literature on graph neural networks.
And finally, we will talk about our contribution, timeSAGE, in adding time sensitive random walks to graph neural networks. We showed on real data with millions of nodes and edges, that our method can
- Generate low dimensional node embeddings.
- Use these embeddings to predict links with high accuracy.
In our context, link predictions are used to identify fraudulent actors by asking questions like, “Are these two devices used by the same person?”, or, “Did the same user create two email ID’s to avail a trial service twice?” etc.
Prerequisite knowledgeBasic concepts of neural networks; mild knowledge of convolutional and recurrent neural networks.
What you'll learn
Deepak Pai is a manager of AI machine learning core services at Adobe, where he manages a team of data scientists and engineers developing core ML services. The services are used by various Adobe Sensei Services that are part of experience cloud. He holds a master’s and bachelor’s degree in computer science from a leading university in India. He’s published papers in top peer-reviewed conferences and have been granted patents.
Carnegie Mellon University
Leave a Comment or Question
Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?
Join the conversation here (requires login)
Premier Diamond Sponsors
Premier Exhibitor Plus
For conference registration information and customer service
For more information on community discounts and trade opportunities with O’Reilly conferences
For information on exhibiting or sponsoring a conference
For media/analyst press inquires