A machine learning system that’s so advanced it’s been used to gain a new understanding of childhood asthma has been made available for everyone to use.
The system is unique because it will create a bespoke learning algorithm for an individual and their data model, so they don’t have to try to create a program using a widely-available standard algorithm that may not be a perfect match. Because the learning algorithm is compiled for a unique dataset, the person who designed it can also understand why the system behaves in a certain way when it’s running.
Infer.NET has been used to try to understand and predict childhood asthma, leading to “the potential of significant clinical impact”. It has also helped to pair gamers playing Halo 5 and match film fans with movies they might like. It is also used in Microsoft services such as Office and Azure.
“Open-sourcing Infer.NET represents the culmination of a long and ambitious journey,” said Yordan Zaykov, a Principal Research Software Engineer Lead at Microsoft Research, Cambridge. “Our team at Microsoft Research embarked on developing the framework in 2004. We’ve learned a lot along the way about making machine learning solutions that are scalable and interpretable.
“Infer.NET was initially envisioned as a research tool and we released it for academic use in 2008. As a result, there have been hundreds of papers published using the framework across a variety of fields, from information retrieval to healthcare. In 2012 Infer.NET won a Patents for Humanity award for aiding research in epidemiology, genetic causes of disease, deforestation and asthma.
“The Infer.NET team is looking forward to engaging with the open-source community in developing and growing the framework further. We have already taken steps towards integration with ML.NET – setting up the repository under the .NET Foundation and moving the package and namespaces to Microsoft.ML.Probabilistic. Infer.NET will extend ML.NET for statistical modelling and online learning.”
Infer.NET can learn as new data is entered into it, which is useful for systems that are used by consumers, and it is also scalable – Microsoft researchers have used it to study information from billions of web pages.