Sites Inria

Version française

Big Data

* - 11/07/2019

Inria at the heart of the HPC-Learning-Big Data convergence

The Teratec Forum is a major event in France and in Europe and the big date in the calendar of international experts in HPC, simulation and big data. The event attracts industrial users, technology suppliers and educational and research institutions. Inria is a partner of the event: join us on 11 and 12 June next at École Polytechnique in Paris.

The theme of high-performance simulation is a key issue for Inria with its links to big data and the increasing importance of AI in high-performance computing. A convergence that is imposing new methods and technologies and is generating the creation of new start-ups. Inria will be presenting some of its research teams specialising in these areas. They will be presenting their latest results, their leading-edge technologies and their start-ups.

Discover the Inria start-ups and talk to our scientists

Automatic analysis of the vascular system and any changes in it is crucial to taking better care of patients and anticipating cardiovascular accidents, which cause 18 million deaths a year worldwide. Existing technologies do not allow surgeons and radiologists to carry out a precise analysis in the time they have available for each patient (less than 30 minutes). Thanks to scientific and technical expertise in image analysis and reconstruction and in integrated mechanical modelling based on a library of computational codes developed by Inria's MEMPHIS team, Nurea offers practitioners software that enables them to reach the necessary level of analysis of each patient's vascular system and the changes in it. The software enables the practitioner to quickly identify the critical areas in the vascular system and to intervene before the accident occurs. For the first time, the diagnosis and monitoring of patients is based on reliable, quantitative information that will be able to reduce cardiovascular accidents.

You will be able to meet Florian Bernard, co-founder and CEO ofNurea at the technical workshop he will be hosting on Wednesday 12 June.

  • The DiagRAMS Technologies start-up: a predictive maintenance solution without adding any new sensors.

With factories becoming increasingly automated, the ability to deal with breakdowns and malfunctions of all kinds, even those that might be less apparent, has become a constant priority for manufacturers. Today, each item of equipment contains a considerable mass of production-related data that is still under-exploited. DiagRAMS Technologies, an Inria start-up, offers a global predictive maintenance solution that does not require the addition of any extra sensors. Our machine learningalgorithms enable us to "monitor" machines' operating status in real time, to anticipate breakdowns and malfunctions using data that already exists in the factory. Our technology is based on the research work of the Modal project team, who are experts in probabilistic statistical learning (supervised or unsupervised) for complex data like raw industrial data: multivariate, missing and partially missing, mixed, high-dimensional functional data, etc. DiagRAMS Technologies is based on the research team's software bricks (Mixmod, Mixtcomp, Blockcluster).

You can meet Margot Correard, co-founder of DiagRAMS Technologies and Christophe Biernacki, manager of the Inria MODAL project team at their technical workshop on Wednesday 12 June.

 

  • TheZettaFlow start-up: a new generic IoT data storage and analysis platform.

Deployable on edge infrastructures and as a cloud service, this platform aims to facilitate the emergence of new real-time data analysis applications that can automate and improve industrial processes on the fly. The lack of real-time visibility over IoT devices and related processes in factories is currently limiting manufacturers' ability to make enlightened decisions based on data analysis. The solutions currently available are costly and ineffective due to the complexity of their multi-layer architecture and static resource management. ZettaFlow will combine decentralised real-time analysis (edge analytics) and "long-term" analyses (cloud analytics) by unifying real-time collection of data, storage of rapid access data and efficient data flow analysis mechanisms. This makes it possible to reduce labour and equipment costs. Overall, ZettaFlow will reduce the effort involved in deploying IoT data analysis systems as well decision-making time, thanks to dynamic automation based on the analysis of data flows via open source technologies. ZettaFlow uses Damaris, a middleware system for efficient I/O management and large-scale in situ data processing, and KerA, a unified system for data flow ingestion and storage. The Damaris and KerA technologies were developed by the Inria KerData project team.

 

  • DEFI, a project team focused on uncertainty quantification methods for digital simulation, optimisation and inverse problems.

The team's approach is centred on the formulation, implementation and validation of digital methods to use computing and available data to run experiments taking account of system uncertainty. It develops methods and related software tools designed specially to meet the challenges posed by manufacturers, for example in the aerospace or energy sectors, and to integrate operational design chains, where necessary.  The final objective is to make the best use of the information from experimental data and digital simulation to enable the optimisation of robust system design in the context of large-scale simulations.

See our European projects

See two European projects in which Inria is a partner at the European Research Café!

  • UTOPIAE, a research and training network looking at cutting edge methods bridging optimisation and uncertainty quantification applied to aerospace systems.
  • BigStorage, a training network that aims to train data scientists in holistic and interdisciplinary approaches around the convergence between HPC and Cloud storage with a view to more efficient management of big data.

As a member of the ETP4HPC (European Technology Platform for High Performance Computing), Inria is contributing, via its researchers involved in different working groups, to the drawing up of the platform's strategic research innovation roadmap.

Inria's teams are therefore contributing to the European research ecosystem in high performance computing and are involved in numerous collaborations in Europe, in particular through their participation in around ten collaborative projects in the Horizon 2020 programme in the HPC field.


The Digital Simulation Awards

Inria is a partner of the fifth edition of the Digital Simulation Awards. A panel made up of leading professionals will will choose the winners of the five awards, which will be presented at the Teratec Forum on 11 June 2019. The Innovation Award will be presented by Jean-Yves Berthou, Director of the Inria Saclay-Ile de France research centre.

Keywords: HPC Big data Machine learning Deep learning Start-up IoT H2020

Top