Joseph Bowles - scalable training of large quantum machine language models

scalable training of large quantum machine language models

Ce séminaire, donné par Joseph Bowles, aura lieu le 16 January 2025, à 15:0. Il aura lieu en salle 24-25/405.

Vous trouverez un plan du campus ici.

Résumé

I will show how quantum circuits comprised of commuting gates allow one to overcome a number of important challenges related to the scaling of quantum machine learning models. For generative machine leaning, I will show how training can be done entirely on classical hardware. This leads to models with thousands of qubits and millions of parameters that can be trained on current hardware, with competitive results to classical machine learning methods. The talk is based on the arxiv papers 2306.14962 and 2501.04776 as well as upcoming work.