As recognized by the Council Recommendation 2009/C 151/02, rare diseases (RD) are a prime example of a research area that can strongly profit from coordination on a European and international scale. RD research should be improved to overcome fragmentation, leading to efficacious use of data and resources, faster scientific progress and competitiveness, and most importantly to decrease unnecessary hardship and prolonged suffering of RD patients. In the specific context of the massive generation, need for reuse and efficient interpretation of data, introduction of omics into care practice and the structuration of RD care centers in European Reference Networks, it appears crucial and timely to maximize the potential of already funded tools and programmes by supporting them further, scaling up, linking, and most importantly, adapting them to the needs of end-users through implementation tests in real settings. Such a concerted effort is necessary to develop a sustainable ecosystem allowing a virtuous circle between RD care, research and medical innovation. To achieve this goal, the European Joint Programme on RD (EJP RD) has two major objectives: (i) To improve the integration, the efficacy, the production and the social impact of research on RD through the development, demonstration and promotion of Europe/world-wide sharing of research and clinical data, materials, processes, knowledge and know-how; (ii) To implement and further develop an efficient model of financial support for all types of research on RD (fundamental, clinical, epidemiological, social, economic, health service) coupled with accelerated exploitation of research results for benefit of patients. To this end, the EJP RD actions will be organized within four major Pillars assisted by the central coordination: (P1): Funding of research; (P2): Coordinated access to data and services; (P3) Capacity building; (P4): Accelerated translation of research projects and improvement outcomes of clinical studies.

AID4SME will facilitate SMEs in developing combined AI and data solutions for large scale resource optimisation challenges for industries that have significant impact on the objectives of the Green Deal. Minimum 20 SMEs, selected through 2 open calls, will receive FSTP to develop these solutions with the support of a Community of Practice (COP). The ambition is to create a COP that will continue after the project lifetime. AID4SME brings together 9 technology blocks and low-TRL playgrounds from 4 scientific partners, to educate and support the SMEs. Additionally, 4 large industry partners (from automotive, whitegoods, battery and energy sector) provide real-life large scale resource optimization challenges that require combined AI and data solutions, and high-TRL playgrounds to integrate and demonstrate the solutions.AID4SME offers an open platform that is flexible to bring in challenges from outside the consortium. AID4SME provides the infrastructure and learning environment that enable the SMEs to solve the challenges, demonstrate solutions and grow into impactful enterprises. The technology blocks cover a wide area of AI & data technologies for the full cycle of data collection, creation of insights, decision support and automation. These technologies have the potential to have significant impact on the Green Transition and boost EU competitiveness for industries. AID4SME will collaborate with the AI-on-Demand platform to enrich its repository with the AID4SME tools and framework, while it is open to deploy the tools/frameworks already available on the AI-on-Demand platform for new use cases. AID4SME will assess the impact of the developed technologies on Green Deal objectives and on social and human aspects. AID4SME brings along partners who are experienced in re-skilling and up-skilling of SMEs and applying standardization as enabler for exploitation. The wide geographical coverage, with partners and DIHs from all across Europe, ensures maximum impact.

Learning continually from non-stationary streams of data is a key feature of natural intelligence, but an unsolved problem in deep learning. Particularly challenging for deep neural networks is the problem of "class-incremental learning", whereby a network must learn to distinguish classes that are not observed together. In deep learning, the default approach to classification is learning discriminative classifiers. This works great in the i.i.d. setting when all classes are available simultaneously, but when new classes must be learned incrementally, successful training of discriminative classifiers depends on workarounds such as storing data or generative replay. In a radical shift of gears, here I propose to instead address class-incremental learning with generative classification. Key advantage is that generative classifiers – unlike discriminative classifiers – do not compare classes during training, but only during inference (i.e., when making a classification decision). As a proof-of-concept, in preliminary work I showed that a naïve implementation of a generative classifier, with a separate variational autoencoder model per class and likelihood estimation through importance sampling, outperforms comparable generative replay methods. To improve the efficiency, scalability, and performance of this generative classifier, I propose four further modifications: (1) move the generative modelling objective from the raw inputs to an intermediate network layer; (2) share the encoder network between classes, but not necessarily the decoder networks; (3) use fewer importance samples for unlikely classes; and (4) make classification decisions hierarchical. This way, during my MSCA fellowship hosted in the group of Prof Tinne Tuytelaars, I hope to develop generative classification into a practical, efficient, and scalable state-of-the-art deep learning method for class-incremental learning.

Microplastic pollution is an issue of increasing concern for the ecosystems and human health. In the last decade, microplastic ingestion and impact in biota has received significant attention. Despite the extensive research conducted, there is still need for improvements in the methodologies used for sample analysis. PlaSeatic aims to provide a new methodology for quantification and characterization of micro- and nanoplastics (MNPs) ingested by organisms in aquatic systems. This method seeks to surpass traditional techniques by reducing the risk of cross-contamination, lowering the size detection limits, preventing alterations and plastic loss while retaining the biological context of the particles. PlaSeatic will focus on copepods, which play a critical role in marine food webs and global geochemical cycles. However, with some adaptations, this methodology could be applied to other organisms or even human tissues. Additionally, this improved methodology could be implemented by other research groups, thereby, increasing homogeneity and accuracy in microplastic research— an urgent need recognized by the scientific community in this field. To achieve the objectives of the project, PlaSeatic rely on the synergy between my expertise in the assessment of ingestion and impact of microplastics in marine food webs and the extensive expertise on chemical engineering, microscopy and spectroscopy of the host (Prof. Roeffaers). This will be further supported by Dr. Pedrotti (supervisor in secondment), whose vast knowledge on plastic pollution and access to historical plankton samples will be invaluable to the project. PlaSeatic will raise public awareness about the impact of plastic pollution. Scientifically, it will enhance our understanding of MNP ingestion, bioaccumulation and impacts of MNPs in zooplankton. It thereby will contribute to the development of biomonitoring and mitigation strategies, and inform policy decisions aimed at ensuring the health of the oceans.