Speaker
Daniel Jaschke
(Istituto Nazionale di Fisica Nucleare)
Description
Tensor-network machine learning is a quantum-inspired method that uses data structures well-known in many-body quantum physics to tackle machine learning tasks. Various ansätze and parameters exist for tensor network algorithms in quantum mechanics, which now can be used as hyperparameters for the quantum-inspired machine learning models. We benchmark hyperparameters, parameters, different Python libraries, e.g., numpy versus torch, and hardware, i.e., CPU versus GPU, to give an intuition for successful and scalable choices amongst the options available in our Quantum TEA "qtealeaves" library.
Primary author
Daniel Jaschke
(Istituto Nazionale di Fisica Nucleare)
Co-authors
Alberto Coppi
(DFA, Univeristy of Padua)
Marco Ballarin
(Istituto Nazionale di Fisica Nucleare)
Simone Montangero
(Istituto Nazionale di Fisica Nucleare)
