OpenIoT & ELC 2016

Today daily objects are becoming connected devices that gather tons of data. The collected data brings little benefits to users, because the devices are programmed to do specific tasks. In order to provide better features, some devices sends the user data to be processed in the cloud. However this won't work in environments with Internet restrictions and can also lead to privacy problems.

The answer for this problem is to have machine learning algorithms running locally. Small IoT devices could delegate the processing to other, more capable, IoT devices in the same environment and under user's control. This way, users will have benefits, without sharing their precious data with third parties.

In this talk we will propose two solutions using Linux and open source libs for fuzzy, neural networks and IoT. We will also compare results, analyze memory and expose its strengths and weaknesses