TECHNOLOGY OVERVIEW

Machine Insight's proprietary technology is GRI (Genetic Rule Induction), a new machine learning algorithm that dramatically outperforms other machine learning algorithms on difficult problems. The prediction models produced by GRI consist of a partitioning of the problem space into distinct regions, each with its own underlying data and predictive capabilities.

GRI has the following performance advantages:

  • GRI can find patterns in data that are invisible to other machine learning algorithms.

  • GRI provides powerful tools for making and controlling predictions on non-stationary time series data. This capability can critically important when making predictions in changing market conditions.

  • Prediction models have clear delineations between what is known and not known. In other words, GRI knows what it doesn't know.

  • GRI is effective on very noisy data.

  • GRI has excellent control over overtraining and data snooping.

  • GRI can handle arbitrarily large skews in predicted outcomes.

  • The predictive models output by GRI are human understandable. Each predictive region is defined in terms of logical combinations of the attribute values, and can be expressed as an SQL query.

  • The predictive models can be updated in real-time, as new data become available.

  • The predictive models generated by GRI are fast, typically 50 microseconds or less per prediction, allowing for predictions to be made in real time.