Neuron™ is a machine learning extension to Oracle Policy Automation. Now you can have the best of both worlds – the traceability of explicit rules and the adaptability of implicit knowledge.
Oracle Policy Automation (OPA) is the world’s leading business rules management system. Business analysts write plain-English business rules into Microsoft Word and Excel and OPA analysis and executed these rules to provide interactive interviews and/or a transactional determination service.
Neuron™ is a 3rd party extension to the core OPA product that adds machine learning capability to the toolset available to the business analyst. This is done by using a set of scenarios with known outcomes for Neuron™ to analyse and in which to detect implicit patterns while training a range of different artificial neural networks. The accuracy of each model is automatically calculated by reference to the same training set and the best model is selected for use in production then converted into a set of mathematical OPA rules that run in the Oracle Determinations Engine together with any explicitly-defined rules.
If an end user is available to confirm or contradict the real-world decisions made by Neuron™, these real-world cases can be fed back into the training set so that the model can be made more accurate or allowed to adapt to emerging patterns while older cases are removed from the set, or have their weighting decreased.
Machine-learning is not suitable to automate decisions that are based on explicit rules, like those found in regulations and policy manuals. For these, the core OPA product is perfectly suited as it provides explanations for every determination is makes, linking the decision back to the user inputs and the business rules. Many decision solutions require a combination of explicit and implicit knowledge, so OPA plus Neuron™ is the winning combination.