Event
Fran莽ois Fouquet, University of Luxembourg
Monday, January 30, 2017 15:30to16:30
Room 3195, Pavillon Andr茅-Aisenstadt, 2920, Chemin de la tour, 5th floor, Montreal, QC, H3T 1J4, CA
Enabling model-driven analytics for cyber聽physical systems.
During this聽talk, I will present what we call model-driven live analytics. I will focus on聽the key enablers to make this approach scalable to the size of country-wide CPSs.聽The main contribution of this work is a multi-dimensional graph data model聽that聽brings raw data, domain knowledge, and machine learning together in a single聽model, which can drive live analytic processes. Firstly, data handled by聽cyber-physical systems is usually dynamic and changes frequently and at聽different聽paces. I will present a temporal graph data model and storage system,聽which consider time as a first-class property and allow to analyse frequently聽changing data. Additionally, I will present how a continuous sequence of sensor聽values can be聽efficiently encoded using live mathematical model inference. Secondly,聽making sustainable decisions requires to anticipate which impacts certain聽actions could have. In some cases, hundreds or thousands of such hypothetical聽actions must be聽explored ahead before any solid decision can be taken. I will聽present our approach to deal with such need - a multi-dimensional model that聽efficiently represent, store and analyse many different alternatives of the聽same system in live. Thirdly,聽to make smart decisions, cyber-physical systems must聽continuously refine behavioural models that are known at design time, with what聽can only be learned from live data. During this talk, I will present how we聽have combined machine learning聽and the multi-dimensional graph data model to聽empower live analytics for cyber-physical systems. Finally, I will conclude聽this talk with details about the open source project, which is developed around聽these research ideas, and share the聽lessons learned with respect to high-performance聽Java and big data.
-----> Cette pr茅sentation sera donn茅e en fran莽ais.聽
-----> Cette pr茅sentation sera donn茅e en fran莽ais.聽