AI-Sprint webinar highlights_March 2021


AI-Sprint webinar highlights



Edge AI is rapidly gaining momentum with growing investments and awareness of its benefits, from reducing costs and latency times for improved user experience to increased levels of security in terms of data privacy through local processing. The momentum around artificial intelligence and edge computing is also demonstrated from an economic point of view, with the AI and edge computing markets expected to be worth $554.3 billion and $250.6 billion respectively by 2024. 

Yet their full potential has yet to be realised. One way to achieve this is by combining various execution platforms for ubiquitous and seamless execution computing environments for a complete cloud continuum. As a result, application developers will have greater control over computing, network and data infrastructures and services while end-users will benefit from seamless access to continuous service environments. AI-SPRINT aims to take full advantage of the potential of artificial intelligence and edge computing by overcoming some of the major challenges in this field. 

The first webinar AI-SPRINT: An EU Perspective on the Future of AI and Edge Computing took place on 30 March 2021 to explain just how AI-SPRINT wants to do just that, zooming in on how it is driving innovations in AI and edge computing as a research and innovation action under Europe’s Horizon 2020 programme. This first in series of online events gave participants the opportunity to learn more about how AI-SPRINT is driving technological innovation, including applications in three compelling real-world cases, outlining key challenges, needs and future trends in artificial intelligence.


Highlights from the AI-Sprint project team

The webinar was hosted by Stephanie Parker, Senior Research Analyst at Trust-IT Services, who introduced and moderated the panel discussion session.

The speakers of the webinar were both members of the AI-SPRINT consortium and external experts.

  • The first speaker was Danilo Ardagna, Associate Professor at Politecnico di Milano and AI-SPRINT scientific coordinator. Danilo opened his presentation by looking at how artificial intelligence systems can be executed in edge computing to enable the analysis of a huge amount of data. When processing large volumes, the computing infrastructure capacity underpinned by cloud and IoT systems becomes more complex due to data heterogeneity, lack of engineering skills, and unstructured development processes. 
    To overcome these issues, AI-SPRINT plans to develop an integrated design runtime framework that provides simplified programming models which lower the time required for developing new AI applications. 
  • The second speaker was Matteo MatteucciProfessor at Politecnico di Milano. Matteo explained the AI-SPRINT project from a systems and infrastructures perspective. The project infrastructure goes beyond the old way of working, where IoT and data were separated from each other. This approach is no longer feasible possible due to higher system complexity, where it is not possible to think about engineering and AI development as totally separated tasks, making the need for centralising data crucial. 
    Challenges like these can be overcome by using a federated learning approach. In this way, novel AI automated design patterns can design new architectures optimised for the cloud computing continuum, reducing resource demand for hyper-parameter tuning, and going beyond AI designed intuition. Moreover, the cloud computing continuum supports the distribution of training algorithms, leverages edge computing resources that are close to data and increases data privacy.
  • Enrico Abate-DagaDirector of Operations at BECK et al, focused his lightning talk with an overview of the three AI-Sprint use cases, chosen to demonstrate the innovations and best practices of the AI-SPRINT framework.
    - The 1st use case deals with healthcare and stroke risk assessment. From a technical point of view, the project develops an edge component, composed of a smartwatch and smartphone acting as a server and another component running in the cloud. The privacy and security of patients' data is guaranteed thanks to the encryption. 
    - The 2nd use case focuses on windmill maintenance, while reducing costs. Windmill maintenance is extremely expensive because of the laborious. processes involved. AI-SPRINT is going to simplify these processes by performing the data analysis as it is collected. From a technical point of view, AI-SPRINT sensors have a camera and the edge server is represented by a computer that moves the data from the edge to the cloud.
    - The 3rd use case centres on farming and the optimisation of phytosanitary treatments in vineyards. The AI-SPRINT goal is to increase treatment precision and, at the same time, reduce the amount of chemicals. AI-SPRINT transforms the tractors into a smart sensor. The amount of chemicals can be tuned based on foliage volumes and shape, humidity and other factors. This guarantees a more precise chemicals distribution in the vineyard. From a technical point of view, the tractor represents the edge and the data inference runs on the edge itself. 


Panel Discussion on challenges and future trends

  • The webinar was followed by an interesting panel discussion which brought  together members of the AI-SPRINT Alliance and project experts for a deep-dive into the challenges, needs and future trends of AI and edge computing from various viewpoints. The external experts invited in the panel discussions were: Atia Cortés, Recognised Researcher at Barcelona Supercomputing CenterDanilo PauTechnical Director, IEEE and ST Fellow at STMicroelectronicsAlfonso Fuggetta, CEO and Scientific Director at Cefriel and Domenico Siracusa, Head of the RiSING research unit at Fondazione Bruno Kessler. Several questions about the future of artificial intelligence and edge computing were asked during the discussion, like: 
Which of the challenges discussed today should we prioritise and why? How do you see AI and edge computing in the context of AI-Sprint?

Federated learning and trainings play a relevant role to overcome the current challenges and guarantee interoperability. Moreover, bringing an industrial and engineering approach to analyse the AI applications can exploit the AI-SPRINT systems at a larger scale.

What difference will cloud computing bring to AI in the context of AI-Sprint?

Cloud computing can bring distributed knowledge across the edge-cloud ecosystem and integration with IoT systems. In this way, the iteration between testing, deploying, distribution and evolution of AI systems can be exploited. Moreover, the cloud provides flexibility and capacity, whenever the edge is saturated. In addition, the cloud can provide a centralised approach, that aggregates the knowledge from IoT.

Considering a lot of new research development on AI and edge computing in Europe and elsewhere. AI-SPRINT is just one example of this, how do you see the landscape evolving in the 2021-2023 timeframe or even up to 2025 when we expect, for example, 5G networks to be widely deployed?

Neural architecture research and hyper parameters optimisation are very important and interconnected to each other as the AI community should not operate without the embedded developers. Cooperation can enrich and simply neural network design and machine learning approaches to increase the overall productivity.


AI-Sprint Alliance

  • The last speaker was Niccolò Zazzeri, Project Manager at Trust-IT Services, who introduced the AI-Sprint Alliance and invited people interested in taking part in it. The main goal of the alliance is to use the AI-Sprint tools and runtime environment to help organisations design solutions which support a better management of the resources efficiency. The Alliance is set up with a dedicated online space, with a dedicated communication channel, a repositories space and collaborative tools to exchange visions, share insights, technological and business innovations. 


Participants and Poll questions

The webinar was well-received with a total of 95 participants, mainly coming from Europe, with a participation rate of 65%. 


During the webinar, some live poll-questions were posed to the audience to better understand the participants’ average engagement and familiarity with artificial intelligence and edge computing topics. The following graphs display the main outcomes pulled out from the short surveys:



Webinar video and slides

Watch the recorded video and download the webinar slides.




About the speakers

Enrico Abate-Daga, Beck et al.

Enrico Abate-Daga is the head of Data to Results Team at Beck et al. He has a Master in Mathematics from the University of Texas. Enrico Abate-Daga has 25 years of professional experience, first as software developer, then as architect and finally as project manager and program manager. In the past five years, he has focused exclusively on Cloud and Big Data projects, in the areas of IT (analysis of logs and monitoring event) and manufacturing (analysis of data from production).

Danilo Ardagna, Politecnico di Milano

Danilo Ardagna is Professor at Politecnico di Milano with a Master degree and PhD in Computer Science from the same university. He was a visiting researcher at IBM TJ Watson Research Center and at the Basque Center for Applied Mathematics. His research interests are performance modelling and cloud/fog systems resource management. He has developed solutions for supporting web services compositions, optimising virtualized systems and advanced schedulers for big data and artificial intelligence applications. 

Atia Cortés, Barcelona Supercomputing Center

Atia Cortés received the M.Sc. and Ph.D. degrees in artificial intelligence from the Universitat Politècnica de Catalunya (UPC), Barcelona, Spain, in 2012 and 2018, respectively. She is currently the Co-Director of the Observatory for Society and AI in the AI4EU platform. She is also a Post-Doctoral Fellow with the Barcelona Supercomputing Center, Barcelona. She has been involved in several research projects funded by the Spanish Council and the EU related to AI solutions for health care.

Alfonso Fuggetta, CEFRIEL

Alfonso Fuggetta, CEO and Scientific Director at CEFRIEL, received a Master degree in Electronic Engineering from Politecnico di Milano in 1982. His working experience spans software consulting and research activities at Politecnico di Milano, as well as collaborating with AIPA, CNIPA, the Department of Innovation of the Italian Government, the Ministry of Health, the Ministry of Labour, the Ministry of Education and of the University. He also developed extensive expertise in software engineering, innovation management, application of ICT to conventional processes and industrial sectors, e-government, and public policies.

Matteo Matteucci, Politecnico di Milano

Matteo Matteucci is Associate Professor at Politecnico di Milano. In 1999 he got a M.Sc degree in Computer Engineering at Politecnico di Milano. In 2002 he got a Master of Science in Knowledge Discovery and Data Mining at Carnegie Mellon University and in 2003 he got a PhD in Computer Engineering and Automation at Politecnico di Milano. Since then, his main research topics are pattern recognition, machine learning, machine perception, robotics, computer vision and signal processing. His main research interest is linked to the development, evaluation and application of techniques for adaptation and learning to autonomous systems interacting with the physical world.  

Danilo Pau, STMicroelectronics

One year before graduating from the Polytechnic University of Milan in 1992, Danilo PAU joined STMicroelectronics, where he worked on HDMAC and MPEG2 video memory reduction, video coding, embedded graphics, and computer vision. Today, his work focuses on developing solutions for deep learning tools and applications. 
Since 2019 Danilo is an IEEE Fellow, serves as Industry Ambassador coordinator for IEEE Region 8 South Europe and Member of the Machine Learning, Deep Learning and AI in the CE (MDA) Technical Stream Committee IEEE Consumer Electronics Society (CESoc).
With over 80 patents, 100 publications, 113 MPEG authored documents and 37 invited talks/seminars at various worldwide Universities and Conferences, Danilo's favourite activity remains mentoring undergraduate students, MSc engineers and PhD students from various universities in Italy, US, France and India.  

Domenico Siracusa, Fondazione Bruno Kessler

Domenico Siracusa (PhD) is the head of the RiSING research unit at Fondazione Bruno Kessler. He is project manager and technical leader of the H2020 EU-Korea DECENTER Project. In the past, he coordinated the H2020 ACINO and EIT Digital DigiFlow projects. He authored more than 100 peer-reviewed publications on cloud computing, networking, security and robustness.