Fairness in AI: Applying deep learning to credit scoring
Who is this presentation for?
- Chief data scientists, chief information officers, chief innovation officers, data scientists, and ML engineers
Machine learning has been used in credit scoring for three decades, making it one of the earliest commercial applications of the technology. However, while machine learning methods have evolved significantly in the interim, with the introduction of elements such as deep learning, credit scoring has never moved on from the use of generalized linear models (GLM). This is largely driven from the need to be able to explain and justify the automated decisions made to both consumers and regulators, which is simple for GLMs but very difficult for newer black box techniques.
Martin Benson details why producing predictions that can be explained and justified is critical to ensuring fair outcomes when machine learning systems are put into production, focusing on credit scoring as a case study. He also examines the challenges that arise in this regard with nonlinear black box techniques such as gradient boosted models and deep neural networks and how these issues were overcome to produce deep neural network models in a unique, highly controlled way that allows their outputs to be understood and justified. You’ll hear about the amazing results that have been obtained by applying Archetype to credit scoring in a fair, controlled, and responsible manner.
What you'll learn
- Discover the importance of explainability in ensuring fairness when applying machine learning and the history of machine learning in credit scoring as a case study
- Explore a system that enables deep neural networks to be produced in a unique, highly controlled way that allows their outputs to be understood and justified
Martin Benson is the head of AI consulting at Jaywing, where he invented the algorithm that underpins the company’s new patent-pending AI product, Archetype, which enables lenders to produces interpretable AI credit-scoring models and is recognized for driving business value from data and developing relevant technology that directly grows business opportunities. He’s led a number of game-changing data science products to discover new business opportunities for Jaywing and its clients, spearheading commercial applications of machine learning, statistical analysis, data mining, and modeling for clients like Avios and Swinton. Martin has a passion for turning data into products, actionable insights, and meaningful stories and is recognized both internally and externally as an expert to solve challenging data problems. He thrives on sharing his knowledge with others and has recently been invited to be an ambassador for the leading deep learning MOOC by Coursera, assisting people in learning about AI. Martin was also a DataIQ Talent Awards finalist for the data science leader category. A trained mathematician with a master’s degree and PhD in mathematics and a leading expert in driving business value from data, Martin is at the forefront of AI and data science.
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)
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