Can a quantum representation reveal useful structure while preserving physical consistency and reliable uncertainty?
The research thesis
My proposed direction combines scientific machine learning with deliberately small quantum models. Candidate problems include classification of simulated physical events, inference of latent parameters and learning compact representations of correlated scientific data.
Physical symmetries, conservation constraints and uncertainty should shape the problem definition. A model that predicts well but violates the underlying physics may be a poor scientific tool.
A first experiment
Use a synthetic dataset generated from a stated physical model, with separate training and test samples. Compare a classical model, a classical kernel and a small quantum feature map under a common tuning budget.
Measure held-out error, calibration, sensitivity to noise and the cost of encoding the data. The experiment should reveal whether any improvement comes from the quantum representation, a different feature choice or additional computational effort.
What a partner would receive
A comparative experiment with documented data generation, model configuration, seeds and evaluation criteria. The immediate aim is a well-founded decision about which representation deserves a larger study.
Scientific connection
This proposed work connects my AI and probabilistic-modeling background with quantum learning research. Published work on the power of data in quantum machine learning makes strong classical comparisons especially important when evaluating a prospective quantum benefit.
Sources & scientific context
Author
Maurizio VivianiIndependent research · Robotics
Take one question into an experiment.
Discuss a focused research collaboration, benchmark or implementation study.
Discuss this direction