Google has been pouring its brain over an Artificial Intelligence project to automate the ‘making sense’ of data sets available across a spectrum of fields – to build an “artificial intelligence for data science”.
The Automatic Statistician project has been brewing under Professor Zoubin Ghahramani and his team at the University of Cambridge who were awarded the Google Focused Research Award for support to their work, reports Kevin P. Murphy, Senior Research Scientist at Google through the Google Research Blog.
Pointing out on the looming issues of Machine Learning, Mr. Murphy explains that inspite tremendous progress made in developing models that can accurately predict future data. problems with ML remain, like firstly the “current Machine Learning (ML) methods still require considerable human expertise in devising appropriate features and models.” Also the output of current methods, however accurate, remains hard to understand and ultimately hard to trust.
Catering to these impending issues, the “automatic statistician”, uses Bayesian model selection strategies to automatically choose good models / features, and to interpret the resulting fit in easy-to-understand ways, in terms of human readable, automatically generated reports.
Complete our SAP x Data Natives CDO Club survey now, and help us to help you
A simplified version as a web-based demo was launched in August 2014 which let the user upload a dataset, and received an automatically produced analysis within minutes. However, now it is being generalized to find patterns in various kinds of data, like multidimensional regression problems, and relational databases. It is reported that early 2015 will see the the release of an expanded version. “We believe this will have many applications for anyone interested in Data Science,” the announcement said.
GigaOm points out that this isn’t the first of its kind attempt towards such technology. Startups like machine learning firm Skytree, BeyondCore, Nutonian and Ayasdi have made or are in the process of making progress with varying degrees of this functionality.
Read more here.
(Image credit: Gwiyddion M. Williams/a>)