The starting step of the solution is to provide the basis system architecture that
enables the implementation of the online machine learning algorithms and visual analytics demanded by the scenario as well as other domains.
PROTEUS will be ready to deal with historical as well as online
data which contribute to the models used for Machine Learning and Analytics tasks. Historical data can serve as basis for the initial prediction and shall be updated during the process with
incoming online data. In addition to the large historical data, PROTEUS will provide a very fast and asynchronous model representation for the real-time analytical/prediction
tasks. The model will be able to asynchronously update as data arrives from sensors over time without having to lock - an operation that is quite expensive when it comes to model training using
tens or hundreds of machines in a cluster. PROTEUS will provide a model management and serving system extending the concept of parameter servers with very fast access without sacrificing fault