A Decision Framework for Machine Learning

Recorded on September 17, 2020

As presented by: Brian Stucky, Senior Enterprise Architect, Quantitative Intelligence, Quicken Loans

Decisions have proven to be a core component of an enterprise digital transformation. Standardization of this capability with the Decision Model and Notation (DMN) provides another key piece to the digital puzzle by facilitating exchange. More importantly, Forrester analysts have noted the importance of digital decisions – along with machine learning and mathematical optimization – as keys to realizing artificial intelligence. DMN gives us the capability for “predictive decision automation” by virtue of enabling the use of predictive models within decisions. However, we can gain even more by using decisions as part of an enterprise machine learning framework.

Machine learning is an extremely powerful technique that allows computer systems to perform a specific task without using explicit instructions. The vast amount of data available – particularly in financial services and mortgage lending – makes this an exciting area of exploration. The potential is enormous, but it doesn’t come without caveats and concerns.

In this presentation we present the concept of a decision framework for enabling machine learning in the enterprise. This framework will be enabled through the use of BPMN and DMN – a process for permissible use, decisions to ensure proper use of data and models, and explainable AI embodied in decisions.

Top of the page