Unsupervised Machine Learning

I believe that unsupervised machine learning is one of the most interesting topics in artificial intelligence right now. It is the only suite of methods which can truly capture major unknowns in data, and thus help with meaningful hypothesis creation.

Currently, I am extending my earlier work on longitudinal unsupervised machine learning to a variety of use cases using a new agglomeration method. This proprietary technology powers my company’s Automated Neural Intelligence Engine (ANIE) and gives some truly remarkable results.

I have been working in unsupervised machine learning since 2007 and have applied methods across a broad range of business problems and datasets. Starting in 2007, I began working with researchers at the Water in Drylands Collaborative Research Program (WIDCORP) in Australia to help understand technology adoption among primary producers in rural Victoria.

We needed to understand the attitudes of producers in the region with respect to climate risk, and to serve that need I created a novel method, termed the Brownell Reduction Method.  The BRM exploits correlations between measurable variables to reduce the dimensionality required to cluster data effectively, given the constraints due to the data collection methodology. It is used to significantly reduce data collection costs in longitudinal research while retaining the ability to track cohesive clusters.  Governments, academic research groups, and Fortune 500 companies use it to efficiently track market trends.

Please see Technology Adoption for a complete list of publications that came from this work.