Skip to main content

AI and Data

Data is everywhere. Technology and manufacturing companies put their data to use, although it is quite a challenge to keep up to speed. Sirris helps you take advantage of the full potential by providing data science and AI support. We are involved in several projects in a variety of domains.

The relevance of data security and data privacy in an industrial context

More and more companies are focusing on how to apply artificial intelligence and machine learning to data, to improve production processes, optimise energy consumption or analyse how their product is used in practice. In this context, data security and privacy is usually considered at a later stage.

In 2020, the annual EluciDATA Tech Talk focused on data security and data privacy, to highlight their importance to the EluciDATA Community. The event took place on 8 December 2020.

Data analysis provides insights into traffic in Brussels during the COVID-19 pandemic

The COVID-19 crisis and the resulting lockdowns have drastically changed everyone’s life and work. For data science, the restrictions during the lockdown have revealed several interesting, real-world phenomena that have proven worth studying. The Sirris EluciDATA Lab took advantage of this opportunity to extract blueprints of the effect on the Brussels traffic.

In early 2020, in the context of a joint industrial doctoral project with Macq and VUB, Sirris EluciDATA Lab started to collect public data of the Brussels traffic, recorded by Brussel Mobiliteit at 55 locations. The objective of the doctoral project sponsored by Innoviris is mainly to develop an advanced trend analysis engine, to provide accurate, situation-aware insights into traffic situations.

In January, we did not suspect that the escalation of the coronavirus pandemic would result in such a unique real-world data set for our research. We have brought together our findings throughout the crisis in a series of blog posts.

EluciDATA Lab involved in research and collaboration on AI

The Sirris EluciDATA Lab joined the Walloon-Brussels TRAIL initiative (Trusted AI Labs), which consists of a general structure made up of universities and research centres, aiming at stimulating joint initiatives and research on artificial intelligence (AI) in Wallonia and Brussels in line with regional policies. In 2020, the EluciDATA Lab helped define the partnership and their services. The EluciDATA Lab is mainly involved in activities related to the adoption of AI in industry.

Three new AI-focused ICON projects

Via the Flemish Policy Plan on Artificial Intelligence, Flanders annually invests €32 million in artificial intelligence (AI). As part of this plan, VLAIO (Flemish Agency for Innovation and Entrepreneurship) launched a special call for artificial intelligence-focused ICON projects (AI-ICON), aimed at bridging the gap between research findings in the field of artificial intelligence and their applications in Flemish industry.

For the special AI-ICON call, experts at the Sirris Data and AI Competence Lab identified three topics with high added value for industry in Flanders, in conjunction with three consortia of academic and industrial partners: TRACY (Trace Analytics), which aims to investigate how to optimally use the log data generated by industrial assets; CONSCIOUS (Contextual aNomaly deteCtIon for cOmplex indUstrial aSsets), focusing on context-aware anomaly detection in industrial machines and processes; and ATWI (Hybrid, multi-modal methodology for Automatic Tool Wear Inspection), which aims to accurately analyse and predict tool wear in metal cutting processes.

Despite the highly competitive VLAIO call all three proposals were selected for funding. Sirris will therefore be able to support the industry further as it tackles real-world challenges related to artificial intelligence.

Project focuses on reliability of electronic components and systems

The reliability of electronic components and systems (ECS) is a major industrial challenge resulting from increased digitisation and the associated complexity of electronic systems in various fields. Unexpected defects or failures can cause economic loss, damage to reputation, and can even result in life-threatening situations. New types of sensors and communications options can collect detailed information to gain insight into the behaviour of these systems, which in turn provides options to enhance the reliability of ECS.

The objective of the iRel4.0 project is to improve the reliability of electronic components and systems along the entire value chain, ranging from chip and packaging to the printed circuit board and system level. The reliability of electronic components and systems needs to be improved faster, so the development processes can be transferred to production more quickly. The quality level also needs to be improved, by means of a fundamental understanding of physical failure mechanisms and the use of Artificial Intelligence methods. Reliability must also be guaranteed when implementing the systems in new, safety-critical and harsh environments, where new materials must be used.

In this project –the joint effort of 75 partners in 13 countries – research into AI and data science techniques and methodologies plays an important role, to gather knowledge for the purposes of forecasting and status monitoring based on data taken from electronic components. These data are used to understand, characterise and predict the behaviour of the individual components and their interactions.

The results will be validated in various application domains, such as the automotive, energy and digital industries. In Belgium, the use case focuses on automotive, more specifically on research into a hybrid methodology to monitor the status of vehicle hydrogen tanks. The Belgian consortium consists of the Sirris EluciDATA Lab, Plastic Omnium, ON Semiconductor, NiniX Technologies and IMEC.

Optimal distribution of AI computing tasks and workloads across existing computing nodes

Recent developments in distributed AI on the edge result in new approaches to secure and targeted distributed analytics. Within Sirris, in the scope of several research projects, we are exploring energy- and resource-efficient scaling of AI-based applications among the existing edge infrastructure, while preserving privacy-sensitive data.

With the advent of edge nodes with increased computational and storage capabilities, companies leverage on these by performing a large range of increasingly resource-demanding applications through a growing number of highly instrumented devices. In the case of industrial monitoring and control applications for example this results in massive amounts of data, distributed across a multitude of devices in the field. Currently, all these data are typically transferred to a central location (e.g. cloud) in view of exploitation by intelligent machine learning and AI.

Truly scalable edge computing  software toolkit

As a decentralised intelligence framework, MIRAI will enable the optimal distribution of AI computing tasks and workloads across existing (often constraint) computing nodes, serving as a truly scalable edge computing software toolkit for IoT and edge-computing applications. Through the MIRAI Framework Building Blocks (MFBB), appropriately sized AI modules will be deployed at nearby available edge nodes. This will provide a low-latency distributed ecosystem for AI-enabled computing in IoT. With application services and tasks deployed on local resources, network problems will become less critical. This decentralised approach will make the MIRAI solution more robust (by enabling new failover mechanisms) and secure (as the computations are executed directly on the source without the need to move the data around).

The ITEA3 project MIRAI was started in 2020 and will run for three years. The Belgian use cases, provided by 3E, Macq and Shayp, focus respectively on distributed renewable energy systems, traffic management and water management.

Shared security solution tackles challenges in connected devices

In recent years, the IoT device revolution has transformed our world into one where everything is connected, smart, and (or should be) secure. Connected devices deliver clear benefits, but they also increase the risk of data manipulation, data theft and cyberattack. The lack of trust by businesses and consumers in smart, connected devices is a barrier.

In this context, the PENTA project SunRISE, which started in 2019 and intends to develop a shared security solution to tackle:
  • machine learning on the edge facilitating IoT security analytics to defend against intrusion attacks and detect anomalies and misconfigurations
  • sharing relevant security data across different stakeholders and applying machine learning techniques on the combined data and models
  • evaluating homomorphic encryption as a privacy enhancing technology and applying machine learning on the combined encrypted datasets.
The Belgian partners Engie Laborelec, NXP and Sirris will demonstrate the SunRISE results in the context of an energy communities (smart grid) use case.