Conference
IEEE ISVLSI 2025 — Quantum neural architectures for wind power
Bologna, Italy
We presented two papers at ISVLSI 2025 in Bologna.
The first compared quantum neural network architectures for wind power prediction, systematically varying the feature map and the ansatz — which, in most QML papers, are chosen once and never revisited. The result was that ansatz depth matters far less than people assume, and feature-map choice matters far more.
The second, with Emine Akpınar and colleagues, applied quantum-enhanced classification to brain-tumour identification from DNA microarray gene expression profiles — a genuinely high-dimensional, low-sample problem, which is where we keep finding hybrid models pull ahead.
The most useful conversations, as usual, happened in the coffee queue.