Quantum@LIACS

Quantum@LIACS 

I am a Professor in quantum computing at the Leiden Institute of Advanced Computer Science, Leiden University and a co-founder and lead of the Applied Quantum Algorithms - Leiden interdepartmental initiative and of the Quantum@LIACS group.

My interests and the research of the group lie in a broad spectrum of quantum computing topics. For the last few years, I have been fascinated with the intersection of quantum computing, machine learning, and artificial intelligence.

In particular, I am interested in:

Quantum-enhanced machine learning

What aspects of machine learning and AI benefit the most from quantum computers?

From low polynomial improvements for various aspects of machine learning, reinforcement learning and AI, to more exotic learning algorithms for many-body physics problems with a potentially BQP-complete flavor and exponential speed-ups, this research line investigates the perspectives and limitations of quantum enhancements for machine learning and AI.

More recently, I have been fascinated with the potential of quantum machine learning to provide the next strides in discovering new physics.

Quantum-applied machine learning

Will advanced AI techniques be indespensible in building scalable quantum computers? 

Modern machine learning and AI methods which can exploit ever-growing amounts of data may  become vital tools to help us come up with new creative and counter-intuitive solutions to problems that arise in quantum computing.

Quantum heuristics

How can quantum computing help solve ubiquitous yet hard computational problems?

Not everything useful has useful lower bounds, and very often, especially in ever more important AI domains, heuristics and in particular quantum heuristic methods will be the only way to go.

Bridging rigorous theory and practice for quantum computers

Quantum computers are proven* to be better in theory. But does this translate to practice?

Most rigorous results on quantum advantage concern asymptotic performance or specific, mathematically characterizable input distributions, whereas practical applications involve  relevant finite sizes and real-world data/input distributions. Meanwhile, although quantum hardware is advancing, many of the most anticipated applications remain out of reach. Can we gain meaningful insights today, without relying on benchmarking that is still infeasible, into whether quantum computers can deliver practical advantages for concrete tasks, realistic inputs, and finite problem sizes while sacrificing as little as possible of the rigor and robustness of the findings?

*-contingent on reasonable complexity-theoretic assumptions