RESEARCH SPOTLIGHT
Hi, I am
Antonino Emanuele Scurria
HOST INSTITUTE
Universite libre de bruxelles
PROJECT TITLE
Advanced training algorithms for neuromorphic photonic information processing
KEY WORDS
physical learning · neural network training · credit assignment · energy-based learning · neuromorphic computing
RESEARCH AIM
My research aims to ground exact gradient-based learning in the natural dynamics of physical systems, so that machines can learn without the costly external computation modern AI depends on.
RESEARCH OVERVIEW
My research combines physics and artificial intelligence. Picture water flowing downhill and settling in the lowest valley without anyone calculating the path—it just follows nature. I study whether a physical object could learn the same way, settling into the “right answer” on its own, instead of through the heavy number-crunching a computer normally does. This matters because training today’s AI burns enormous energy, much like running a huge factory. The brain, by contrast, is more like that flowing water: efficient and effortless. The goal is hardware that learns by letting physics do the work.
RESEARCH CHALLENGE
A key challenge in my research is enabling exact gradient-based credit assignment using only the local dynamics of a real, physically realizable system, rather than relying on backpropagation's non-local backward pass.
RESEARCH INNOVATION
What makes my approach unique is that, unlike modern AI, which runs learning as software on power-hungry hardware, my research explores how the physics of the hardware itself can perform the learning, making efficiency an inherent part of the system.
RESEARCH IMPACT
What excites me most is the idea that learning—something we think of as abstract computation—might actually be a natural physical process, the same way a system settling into equilibrium is. It blurs the line between physics and intelligence.
RESEARCH SUMMARY
My research connects physics, machine learning, and computing hardware. The central question I explore is whether learning can emerge from the intrinsic dynamics of a physical system, instead of running as an abstract algorithm on standard digital processors. The reason matters in practice: neural networks today are trained on power-hungry hardware using procedures with no real physical counterpart, whereas physical, neuromorphic, and photonic systems could in principle learn on their own, with much lower energy use and at speeds set by the physics itself. My work explores how learning might be grounded in basic physical principles and how such learning rules could apply to realistic systems that are non-conservative and decentralized. My wider goal is to narrow the gap between computation and the physics that runs it, toward efficient, brain-inspired hardware that learns through natural physical processes.
RESEARCH VISUAL
RESEARCH JOURNEY
RESEARCH SECONDMENT
Akhetonics, Germany (2 months) & CSIC, Spain (2 months)
During my secondments at Akhetonics, Germany, and CSIC, Spain, I will broaden both my theoretical and practical understanding of optical computing. At Akhetonics, I will work on theoretical challenges related to optical computing, with a particular focus on logic synthesis, while gaining valuable insight into real-world hardware design and implementation. At CSIC, I will receive training in wavelength-division multiplexing (WDM) optical computing systems, strengthening my understanding of analytical techniques and methodologies for studying complex systems.
LEARNING GOALS
My goal is to bridge the gap between theory and application. I look forward to exploring this problem through a dual lens: leveraging industrial R&D perspectives on one hand, and collaborating with a different laboratory to integrate their distinct theoretical approach and expertise on the other.
RESEARCH BENEFITS
These collaborations will significantly strengthen my project. First, the industrial partner will provide a practical perspective focused on production challenges and real-world problem-solving. Second, collaborating with the other lab—which approaches the same theoretical question from a different angle—will equip me with a broader set of technical and analytical tools.
CAREER DEVELOPMENT
In terms of my future career, this experience will make me a highly versatile professional. By working at the intersection of academia and industry, I will develop both the rigorous mindset of a researcher and the pragmatic, results-oriented focus of a developer. This dual background is exactly what top-tier R&D departments look for, as it allows me to bridge the gap between complex science and commercial reality.
MORE THAN PHD
Why did you decide to join POSTDIGITAL+ ?
I joined POSTDIGITAL+ because it sits exactly at the intersection I care about—physics-driven learning and the hardware that could make it real. It connects the theory I work on to neuromorphic and photonic platforms where these ideas can actually be tested.
What do you think is the biggest advantage of being part of the International Doctoral Network?
The biggest advantage is developing a globally adaptable, industry-ready mindset by learning to translate complex research into real-world solutions across diverse perspectives and laboratories.
What skill have you developed the most so far?
I’ve grown most in connecting abstract theory to physical reality—learning to ask not just ‘is this correct?’ but ‘could this actually run on a real device?’ That shift in thinking has reshaped how I approach problems.
What are your career ambitions after completing your doctorate?
I want to stay at the interface between fundamental research and real-world technology. Whether in academia or a startup, my goal is to help translate physics-driven learning into computing that is genuinely more efficient—and to keep working across the disciplines and people that make that possible.
What inspired you to pursue this research Area?
What inspired me was a simple discomfort: the way AI learns has no counterpart in nature. Coming from physics and mathematics, I wanted to understand whether learning could be described by the same principles that govern everything else—and that question pulled me in.
What surprised you most since starting your PhD?
What surprised me most is the mix of talent, openness, and diversity. The researchers across POSTDIGITAL+ are genuinely excellent and remarkably approachable, and they come from very different countries, disciplines, and backgrounds—which makes every discussion richer and more unexpected.