Neural networks a new Google project
According to some authors an artificial neural network (ANN) is a network inspired by biological neural networks(the central nervous systems of animals, in particular the brain) which are used to estimate or approximate functions that can depend on a large number of inputs that are generally unknown.
Jim Gao, engineer equipment data centers Google has been working on machine learning, particularly the issue of neural networks, and has managed to significantly increase the already high energy efficiency of data centers Google using for their predictive models’ ability to generate datacenters and now auto analyze their efficiency ratings.
The variables that influence these efficiency ratios related data such as the temperature of the servers, the outside air and the demand from the IT, being recorded every 30 seconds and summarized in a general standard measure called PUE Power Usage Effectiveness, a relationship between energy expenditure and its actual use in IT. Well, what he has accomplished this genius named Jim Gao is to establish models that predict for energy efficiency improvement after certain actions, models that have achieved up to 99.6% effectiveness.
For more information, especially for those working in areas related to data centers and must deal daily with the cooling of the server infrastructure and optimization of energy efficiency, it has provided a full article titled Machine Learning Applications for Data Center Optimization (PDF) that explains in detail the methodology, statistical basis, real examples of what has been achieved and finally some conclusions, limitations and additional references.
Savings and money stand in the official Google blog is representing the important work of Jim Gao which, like Gmail and other major products of the company, has been achieved thanks to the use of 20% of the time provided by Google to all employees to invest it in personal ideas, only the remaining 80% in developing ideas for the company.