BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Nathan Wiebe (Pacific Northwest National Labs\, University of Wash ington) DTSTART:20200529T000000Z DTEND:20200529T010000Z DTSTAMP:20260423T005752Z UID:UTSQSI/2 DESCRIPTION:Title: T raining fully quantum Boltzmann machines\nby Nathan Wiebe (Pacific Nor thwest National Labs\, University of Washington) as part of Centre for Qua ntum Software and Information Seminar Series\n\n\nAbstract\nIn recent year s quantum machine learning has grown by leaps and bounds but a major probl em still vexes the field is how to efficiently train quantum neural networ ks. This is particularly challenging because of the lack of a natural bac kpropagation algorithm for updating the quantum model. \nIn this talk\, I will focus on an approach that can mitigate this problem through generativ e training. We will show how to construct a fully quantum model of a Bolt zmann machine and train all of the parameters of that model for both the q uantum and classical parameters in the model. In contrast\, existing meth ods were not able to achieve this. \nIn particular\, we will show explicit query upper bounds for the cost of simulation\, provide a formal proof fo r BQP-completeness for evaluating such neural networks and also discuss re maining problems in the field and how to generalize the ideas presented he re to go beyond Boltzmann machines to allow efficient training of broad cl asses of quantum neural networks.\n LOCATION:https://researchseminars.org/talk/UTSQSI/2/ END:VEVENT END:VCALENDAR