Celebrations for ICTP's 60th anniversary are taking place throughout 2024 and around the globe, spreading the excitement of this momentous year to the Centre's considerable global network. From public science talks to scientific conferences and workshops, ICTP is using its anniversary as an opportunity to promote sustainable science with a view to the future.
Recently, ICTP held two anniversary science events in Nepal and Macau, both of which focussed on technological innovation. In Nepal, the School on Parallel Programming and Parallel Architecture for High Performance Computing, which ran from 22 April to 3 May 2024 and was co-organised by long-time ICTP collaborator Tribhuvan University, aimed to train early career scientists on tools and techniques to help them take full advantage of scientific computing for their research projects. More and more scientists are turning to the power of parallel computing and Machine Learning (ML) techniques to assist them with simulations and data analysis, adding an edge to their research by using tools that only a generation ago were unavailable.
These tools, however, require new skills in programming, and that was the focus of the School. Through a combination of lectures and hands on sessions, the 40 participants learned fundamental techniques in parallel programming, GPU programming and Machine Learning, all fundamental to harnessing the full power of high performance computing resources.
"This was the first school on high performance computing ever to be held in Nepal," said Ivan Girotto, HPC Applications Specialist and a co-organiser of the activity. Participants took full advantage of the high-level training and expertise of the instructors to work on independent projects for which they could apply their new skills; these included optimization of post-processing analysis code for data analysis on chemical reactions, and application of ML on satellite data. "During the school, participants had access to world-class high performance computing facilities on the Leonardo supercomputer hosted at CINECA, thanks to the HPC agreement with ICTP," explained Girotto, adding that the access continues for two months after the school (or longer by request).
ICTP teamed up with the United Nations University Institute in Macau for a workshop on TinyML for Sustainable Development. During a one-week session from 26 to 30 April 2024, some 18 participants from 11 Asian countries learned how TinyML, a new technology that allows Machine Learning models to run on low-power microcontrollers, could be used to develop applications in resource-constrained environments. Participants learned how these tools can have multiple, practical applications, from monitoring air pollution to tracking weather conditions for agriculture. Lectures on programming and hardware were complemented by hands-on sessions where participants built and programmed their own devices.
According to Marco Zennaro, head of ICTP's Science, Technology and Innovation unit and the event's co-organiser, the workshop focused on applications that are particularly relevant to Asian researchers. "We had discussions on Asian issues that can be tackled by TinyML. As examples, participants spoke about environmental issues, building-stability monitoring, e-Health, just to name a few," he explained.
Participants were encouraged to apply their knowledge from the workshop's sessions to the development of a project. "One of the most interesting project presentations was by a participant from Bangladesh, who proposed the development of biosensors for shrimp pathogen monitoring using TinyML," recalled Zennaro, adding, "The project encompassed all of the benefits of TinyML: it applied AI to solve a local issue, it used the offline capabilities of TinyML (no Internet required), and it can be scaled, as the solution is low cost."
Zennaro said that given its success, ICTP and the UNU Institute are considering making the TinyML workshop a yearly event in Macau for the Asian region. That would mean spreading knowledge on this cutting-edge technology in a part of the world hardest hit by climate change and extreme weather. Workshop participants become part of ICTP's TinyML Academic Network that now includes more than 50 universities in developing countries. "We organize online workshops, seminars and have an online network of over 1000 researchers in TinyML," explained Zennaro.
This small but growing community of trained specialists could play a significant role in helping their countries achieve the UN's Sustainable Development Goals and in enabling new applications in fields such as healthcare, agriculture, environmental monitoring and conservation.
Related links: