AI4EOSC
stands for Artificial Intelligence for the European Open Science Cloud
Artificial Intelligence (AI) along with Deep Learning (DL) and Machine Learning (ML) are at the forefront of scientific and industrial research. The impact of these techniques, together with the avalanche of large datasets in the big data era is transforming science and innovation, opening many new research areas.
[
Expectations are high and so is their potential
Besides, machine and deep learning applications and services are increasingly more demanded by research and innovation stakeholders as pathways to effectively build and exploit tools based on these techniques.
The vision of the AI4EOSC project is to increase the service offer in the EU landscape by expanding the European Open Science Cloud (EOSC) ecosystem to support the effective utilization of state-of-the-art AI techniques by the research community.
In this regard, the project will provide highly innovative services built on top of existing EOSC services, thus allowing EU researchers to efficiently exploit large and distributed datasets, following a service-oriented approach over the EOSC continuum.
The AI4EOSC project bases its activities on the technological framework delivered by the DEEP-Hybrid-DataCloud H2020 project, allowing researchers to exploit computing resources from pan-European e-Infrastructures.
The AI4EOSC platform is a production-ready system that is being effectively used by researchers in the EU to train and develop machine learning and deep learning models.
AI4EOSC is continuously enhancing this platform, delivering new high-level services and functionalities, targeting direct exploitation by scientific teams, allowing them to reduce the time to results and increase productivity by building better analytics tools, products, and services leveraging artificial intelligence, machine learning, and deep learning, with focus on advanced features like federated learning, split learning or distributed training. AI4EOSC makes a special emphasis in ensuring that all the research outputs and sub-products (data, models, metadata, publications, etc.) adhere to the FAIR data and research principles.
You can check our project deliverables and publications in our dedicated page.
Goals
The main goal of the AI4EOSC project is to foster an AI, ML and DL exchange in the EOSC context, enhancing the current EOSC service offer delivering added value, innovative and easily customisable services serving a broad range of scientific users.
As such, we will focus on tools to provide AI, ML and DL services by integrating into the project real life use cases to co-design the project proposal and drive our integration activities.
This overarching goal will be set up by fulfilling the following project specific objectives:
[01]
Providing ML practitioners with feature-rich services to build and deploy customisable ML, DL and AI applications following a platform and serverless approach with horizontal scalability over the EOSC continuum
[02]
Enhancing existing cloud services to support ML and DL on distributed datasets, with a particular focus on federated learning. Delivering methods to build and compose machine learning and deep learning tools, making possible the development of more complex data-driven composite AI applications.
[03]
Fostering a ML and DL exchange in the context of the European Open Science Cloud, enhancing and increasing the application offer currently available in the DEEP Open Catalogue.
[04]
Extending the service offer and the capabilities available through the EOSC portal, coordinating with the operational and management activities carried out by existing and future initiatives, creating and establishing cooperation synergies whenever possible.
Cases
In order to investigate the effectiveness and feasibility of the AI4EOSC project concept and approach, we’ve selected three particular use cases from several different real-world ones that originated from two distinct scientific disciplines, each having a direct business impact. The scientific domains covered by these use cases highlight the interdisciplinary nature of the use cases.
The use of cutting-edge technologies, such as deep learning, federated learning and composite Artificial Intelligence, demonstrates the potential of these use cases to drive innovation and bring significant benefits to the target users. The requirements of the use cases will also steer the AI4EOSC platform development.
Moreover, the project consortium will make an active effort to maximise project impact by onboarding of external use cases.
An exciting venture to improve energy efficiency in urban settings through the power of AI and thermography.
[Automated
Thermography]
The solution uses Deep Learning models to detect hotspots through instance segmentation in combined thermal and RGB image data.
The technology will be hosted on a cloud-based automated service that leverages best practices and technology advances of the AI4EOSC platform, potentially using decentralised learning techniques such as federated learning by selecting each client according to the geographic location where an image was taken.
The target audience for this use case includes urban planners, building owners, and district heating network operators who struggle to maintain high energy efficiency due to the inability to quickly and accurately pinpoint the location of heat loss. The goal is to automate the detection of thermographically salient heat losses to accelerate the implementation of necessary countermeasures and repairs to mitigate their effects.
Join us in AI4EOSC project and stay one step ahead of thunderstorms!
[Agrometeorology]
AI4EOSC uses the power of AI and meteorological data to provide timely and precise warnings for farmers and local communities. By combining radar imagery, in-situ measurements and numerical weather predictions, our project generates added-value products that improve farmers’ activity and minimise the damage caused by high-impact weather conditions related to thunderstorms such as squalls, hail, lightning or flash floods.
Our AI-based approach enables us to grasp complex non-linear features of natural processes. Particularly, it combines large-scale data with local point measurements to benefit from the preciseness of the ground based data and the spatial coverage from the radar imagery.
Our target users include farmers, public administration, and local governments. With our thunderstorm warnings, farmers can plan their work activities and ensure their safety, while public administration and local governments can prepare emergency response plans and minimise the impact.
Join us on this mission to revolutionise disease detection in agriculture and improve the sustainability of our food production systems.
[Integrated
Plant Protection]
This use case aims to revolutionise plant disease detection methods. Our goal is to augment currently used mathematical models with the power of AI-based models, developed and scaled on the AI4EOSC platform. Our AI-based solution combines a network of meteorological data, existing mathematical models, and ground observations, enhanced with satellite data, to provide greater terrain coverage and spatial precision.
Using AI4EOSC’s federated learning and composite AI solutions, we are pushing the boundaries of disease detection models in agriculture, which will be integrated into existing national advisory platforms, such as eDWIN, operated by AI4EOSC partners (WODR and PSNC), and accessible to farmers, advisors, and scientific institutes. We offer individual risks of cumulative risk calculations for the most common crops and related diseases including potato blight and Cercospora in beet.
With a target audience of farmers, public administration, local governments, and institutions responsible for monitoring hazards in agriculture, we plan to reach around 100,000 users in Poland alone, with the potential for scalability to other countries and platforms. Our ultimate goal is to improve the quality and safety of food production by reducing the usage of pesticides.
Partners
