IRIXYS Research program and scientific agenda:
For many years now, the IRIXYS scientific agenda has been structured around three research subjects:
- Cognitive Computing, Big Data Analytics and Information Retrieval
- Cybersecurity and private life protection
- Federated collaborative systems (Cloud-Edge-Fog-Computing, Blockchain, Internet of Objects, Industry 4.0)
The research programs linked to at least one of these priorities, or sharing an interface with these themes are prioritized. We are using the complementarity of our skills to bring an answer to the most complicated scientific stakes and questions. To do so, we are combining different opinions and approachs, and rely on complementary tools and methodologies, working around the blurry line between scientific fields, where problems are both the most complicated and impactful ones.
This common research agenda and approach provides operational and scientific framework of all co-supervised PhD thesis, publications and common projects.
The list of all workshops, as well as their themes, can be found under the Academic Exchanges page (in the drop down menu). The first workshop dates back to 14 years ago.
In 2021 and 2022, the IRIXYS scientific team led a profound reflexion on the future orientation and development of our research program. It was both based on our own scientific analysis but also on regional, national and European initiatives and strategies.
In each of these themas, some lines of work have strengthened, or more recently, emerged. Some examples could be the notions of explainability and confidence in machine learning or blockchain and Web 3 in federated IT. The multi-skilled IRIXYS team also greatly contributed in the concept of “federated learning”, interlinked between machine learning and private life protection. The knowledge IRIXYS had in both these fields made a major difference.
We aim at defining three axes of transversal works:
- Environmental impact, green IT
- Ethics, trust and protection of private life
- Numerical sovereignty and open science
Even it they define specific disciplinary fields, these dimensions are cross-cutting by nature, We thus aim to consider them in all of our research field by:
- Developing methods and approaches that minimise the environmental impact
- Reinforcing our work on privacy, explainability and empowerment of the user – These are three of the mandatory conditions to guarantee a trustful and responsible numerical society.
- Still placing a strong part of our work in the decentralised systems general framework: We have been defending its potential since the founding of MDPS/IRIXYS, to develop performing, extensive, sovereign and protective solutions for private life.
- Contributing to open science initiatives.
More precisely, in the framework of our work in cognitive computing and machine learning, we focus on:
- Explainability, interpretability and causality
- Decentralisation and privacy, federated learning
- Fairness, diversity, bias minimisation
- Knowledge representation, visualisation and interactivity
- Machine learning at the edge
- Treatment of very big volumes of data and knowledge
This thema is fully embedded in the scientific strategic agendas of our institutions. Therefore, in Passau, and thanks to the subvention granted within the High-Tech Agenda Bayern, chairs have been formed in the methodology and ML application fields. The themes in this context go from analysis and prediction in Big Data to analysis and prediction in explainable AI. For the latter, the main goal is to create a path to efficient usage of complex ML algorithms in practice. Automatic treatment of natural languages forms a common application field. In Lyon, the structuring project in the “easely” engineering (EASELY: Engineering at Saint-Etienne Lyon), includes a “numerical society” component and a thematic program named “Data and AI: from production to governance in a durable and responsible approach”.
As for the decentralised and collaborative distributed systems, our research works range from a theoretical aspect (e.d distributed algorithms (consensus)) but also on a more systemic level, including questions of error tolerance, resilience and selfishness (in this field, we benefit from the expertise of Milan University in game theory) at the interface of our cybersecurity works (see below). Many architectural goals are of interest to us: cloud computing, edge/fog computing, P2P networks, blockchains and distributed ledgers. In these last few years, we have strongly invested ourselves in this latter axis. Notably, with Lyon, a common laboratory was created, co-financed by the ANR, with the iExec company, strongly involved in the blockchain eco-system. Thus, iExec has welcomed several interns from the PhD-Track
, and is the host of 2 doctoral students at the moment(CIFRE convention and ANR doctoral contract). Blockchains, by nature, guarantees properties of immutability, non repudiability/traceability, of disponibility/fiability and a lack of third party trustfulness (in the case of public blockchains). Through all this, it completely renews research in federated systems, and offers brand new opportunities. It does, however, raise difficult problematics in the process. We will keep on working around this theme (consensus algorithms with low carbon impact, based on reputation, interchain and hybrid protocols, applications based on a blockchain infrastructure…). We also aim to strongly develop our activity around Web 3, that offers a strongly decentralised vision based on the blockchain technology. It aims at answering to sovereignty, centralisation and private data protection problematics that the Internet is raising.
The Internet of object is also a field that we had studied a lot in the last few years. In collaboration with Siemens and Atos, we have been able to drive common works aiming to optimise and better the treatment of data coming from sensor networks. Based on this expertise, we have also directed a UFA work group on 4.0 industry and standardisation.
With our industrial partners, we wish to intensify those works and to explore them in multi-scaled architectures (“Industrial Edge/Fog”). This is done by crossing them with security requirements, personal data protection, and minimisation of environmental impact.
We are also developing, since the IFIK/MDPS formation, an intensive research activity in the cybersecurity and personal data protection fields, particularly within co-supervised works, often linked with industrial applications. This includes IoT, smart transportation (helped by Lyon’s expertise), 4.0 enterprise (Milan’s expertise); but also standards integration and development (Passau’s expertise) and certification mechanisms (Milan’s expertise). We aim at reinforcing our fundamental works, notably in post quantic security, started in Milan, in the the secured multi-party computation (SMPC). It is one of the paradigm that we prioritise, due to the theoretical guarantees in brings inside IFIK/MDPS. Regarding privacy, for the last 10 years, we have prioritized the smart city domain, including in mobility, with the help of LabEx, on Urban World Intelligence, whose dimension of a “smart city” was carried by researchers from IFIK/MDPS. While pursuing our activity in this field, offering problematics that are scientifically very interesting, and whose societal impact is naturally very significant, we aim at broadening our field of investigation. This is done within projects such as Web 3 (iExec partnership), OpenWebSearch (see below), autonomous vehicles (a co-supervised thesis is currently happening; W3C partnership), recommandation systems (several thesis running) and machine learning (federated learning (several thesis running) privacy-preserving machine learning at the edge).
Within this context, we rely methodologically on the adapatation and usage of cryptographic techniques (zero knowledge disclosure proofs, homomorphic encryption, hash functions…), of multi-party secured computing (SMPC) and of statistical perturbation mechanisms. These methodological approaches are powerful and provide a mathematically mastered methodological framework, allowing us to derive from formal guarantees.
At the intersection from these three core thematics within the scientific project IFIK/MDPS, we also aim at opening a new axis of federative research “Open Web Science / Numerical Data Sovereignty) in order to contribute to the development of a numerically sovereign Europe. With the help of the OpenWebSearch
project, coordinated by Passau and where part of the IFIK/MDPS research team is involved, we aim at working for the creation of infratructures and open, unbiased research engines, that are in line with the european rights and values.
Within the PhD-Track research program, we will focus ourselves on modeling and creating indexes and search engines. In parallel, we will bring our expertise to the data protection field and will study how efficiency, data protection and transparency can be conjointly tackled.
The problems approached here, just like within the Web 3 framework, are raising questions far further than just computing: law, economy/data economy, company management, psycho-sociology. An interdisciplinary approach is thus mandatory in order to consider all of its expectations and consequences. Exchanges already exist with our colleagues from other faculties within our universities, notably the law faculty of Passau and Milan, and with the Gaston Berger Institute of Lyon. These links will be strengthened with time.
In parallel, Univerity Passau is coordinating a project of european university network (EUN), where INSA Lyon is very actively part of the project. One of its three strategic axis is called “Sustainable AI and digital technologies in Europe”. This project, perfectly phased with the IFIK/MDPS scientific project, relies on a consortium involving SHS specialized universities. IFIK/MDPS, including its researchers, doctoral students and students, will be able to clearly benefit from this initiative and from the interdisciplinary dynamic that it aims to create.
Finally, as we mentioned above, the numeric represents a significant and ever growing part of global electric consumption these days. Taking this environmental impact and IT applications into account can not be a side consideration anymore. It is mandatory to turn computing into a more efficient and environmentally respectful tool. Within our research program framework, we are eager to place this requirement at the core of our work. It is already integrated in multiple scientific programs of ours. This evolution includes the development of metrics to measure and foresee the energic consumption of our solutions. A calibration of these metrics will be carried out in our own infrastructures. Preparatory works have already begun in Passau and Lyon. We will also rely on the “responsible numeric” teaching chair, currently being formed in the INSA IT department, that tightly associates researchers, students and companies. The PhD-Track students will naturally have a training rôle in the project and the partnerships created with the companies will nourish researches conducted within the PhD Track.
Our 2023-2027 agenda conveys both the will to deepen our reference research axes and to open our research program on rising questions such as numerical sovereignty, environmental impact or open science. This will be done by prioritising large scale projects associating academic and socio-economic partners.
In this context, we want to highlight that in may 2023, our 30th scientific workshop will happen in Lyon. For this occasion, a French-German day (in partnership between the UFA KarlINSA integrated curriculum) will happen on May 12th. We will rely on our bilateral programs, such as IFIK/MDPS, IRIXYS and KarlINSA, to introduce and illustrate (in a interdisciplinary approach) the thme of Open Science. The first concepts of an Open Web Index will be shown to the public.
In december 2023 in Passau, we will position University Passau as an intergenerational and interdisciplinary research center in the digitising and sustainable development fields. The first concepts of a green data centre settlement will be unveiled to the public and the French-German dimension of it will be discussed. Both workshop agendas will be tightly linked to the UFA/DFH activities for the Elysée Treaty’s 60th anniversary.
Links with companies:
A very prowerful link is held with the economic actors, great multinational companies (Atos, Worldline, Airbus, Siemens, etc.) or innovative and regional SMEs (Lyon: AttestationLégale, WebCastor, SIS, iExec, La French Tech…; Passau: One Logic, Innkubator, DB-Regio…; Milan: Engineering SpA…), our three regions (Auvergne-Rhône-Alpes, Bavaria, Lombardy (the University of Milan being a partner of the international doctoral college MDPS and of the IRIXYS centre)), being among the main European economic driving force.
This link comes to play (with the IFIK-MDPS students in master’s or PhD degrees) in different ways:
- A very large majority of the thesis subjects and thesis of Masterarbeit (except some very prospective and fundamental subjects) come from partnerships involving industrialists (thesis in CIFRE convention, projects carried out in consortium, contracts, etc…). Despite the fact that it brings students a scientific contect both in line with concrete problematics and powered by real systems (which allows, notably, to have experimentative data, mandatory to validate the proposed approaches), this partnership with socio-economic actors allows the students to develop their network and their perception of industrial and economic stakes, as well as their understanding of companies’ realities.
- In almost all of IRIXYS scientific workshops, industrialists take part in the exchanges, bring their feedbacks to doctoral students, sometimes “challenge” them, and discuss with them out of the sessions. These feedbacks, integrated in the thesis carried out in CIFRE conventions, are of great value for all doctoral students, because it brings them a different view from the academical point of view from their tutors, while questioning their results and practices. The participation of industrialists is also contributing to the enrichment of their network.
- Finally, as mentioned above, IFIK/MDPS students benefit from tools, helping them entering into profesionnal life, offered by the IF department and the LIRIS laboratory: simulation of interviews with HRDs, “industrial days” of the laboratory and “industrial meetings” of the Department.