Speakers
Mr
Rohit Sarma Sarkar
(Universitat Politècnica de Catalunya)
Rupayan Bhattacharjee
(Universitat Politècnica de Catalunya)
Description
Quantum Signal Processing (QSP) generates univariate polynomials but is limited by high circuit depth and parameter cost owing to its parity constraints. We introduce Weighted QSP (WQSP), where a weight function modifies the signal operator, recovering QSP as the unit-weight case. Non-trivial integer weights yield a pruned QSP with significant reduction of parameter count vis-a-vis circuit depth, achieving competitive polynomial approximation error with fewer resources. Generic weights admit error bounds and can yield exact approximation in many cases. We analyze WQSP from deterministic and quantum neural network viewpoints, showing it retains QSP-level expressivity with improved efficiency.
| Topic | Quantum algorithms for machine learning applications |
|---|---|
| Additional Topic | Theoretical foundations of quantum learning |
Authors
Mr
Rohit Sarma Sarkar
(Universitat Politècnica de Catalunya)
Rupayan Bhattacharjee
(Universitat Politècnica de Catalunya)
Co-authors
Elías Combarro
(University of Oviedo (Spain))
Michele Grossi
(CERN)
Lirandë Pira
(NUS)
Dr
Carmen G. Almudéver
(Universitat Politècnica de Valencia)
Dr
Sergi Abadal
(Universitat Politècnica de Catalunya)
Eduard Alarcon
(UPC BarcelonaTech)