BEGIN:VCALENDAR
VERSION:2.0
PRODID:www.dresden-science-calendar.de
METHOD:PUBLISH
CALSCALE:GREGORIAN
X-MICROSOFT-CALSCALE:GREGORIAN
X-WR-TIMEZONE:Europe/Berlin
BEGIN:VTIMEZONE
TZID:Europe/Berlin
X-LIC-LOCATION:Europe/Berlin
BEGIN:DAYLIGHT
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
DTSTART:19810329T030000
RRULE:FREQ=YEARLY;INTERVAL=1;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
DTSTART:19961027T030000
RRULE:FREQ=YEARLY;INTERVAL=1;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:DSC-22786
DTSTART;TZID=Europe/Berlin:20260330T153000
SEQUENCE:1774848953
TRANSP:OPAQUE
DTEND;TZID=Europe/Berlin:20260330T163000
URL:https://dresden-science-calendar.de/calendar/en/detail/22786
LOCATION:MPI-PKS\, Nöthnitzer Straße 3801187 Dresden
SUMMARY:Wanjura: Physical learning machines
CLASS:PUBLIC
DESCRIPTION:Speaker: Dr Clara Wanjura\nInstitute of Speaker: MPL\nTopics:\n
 Physik\n Location:\n  Name: MPI-PKS ()\n  Street: Nöthnitzer Straße 38\n
   City: 01187 Dresden\n  Phone: + 49 (0)351 871 0\n  Fax: \nDescription: T
 he increasing size of neural networks for deep learning applications and t
 heir energy consumption create a need for alternative more efficient hardw
 are. The field of neuromorphic computing aims to address this challenge by
  replacing our digital artificial neural networks with a physical network\
 , for example\, using optics\, that performs the required mathematical ope
 rations. Current proposals and implementations rely on physical nonlineari
 ties or optoelectronic conversion to realise the required nonlinear activa
 tion function. However\, there are considerable challenges with these appr
 oaches related to power levels\, control\, energy efficiency and delays.  
  In the first part of my talk\, I will present a scheme [1] for a physical
  neural network that relies on linear wave scattering and yet achieves non
 linear processing with high expressivity. The key idea is to encode the in
 put in physical parameters that affect the scattering processes. Moreover\
 , we show that gradients needed for training can be directly measured in s
 cattering experiments. We propose an implementation using integrated photo
 nics based on racetrack resonators\, which achieves high connectivity with
  a minimal number of waveguide crossings. Our work introduces an easily im
 plementable approach to neuromorphic computing that can be widely applied 
 in existing state-of-the-art scalable platforms\, such as optics\, microwa
 ve and electrical circuits.  In the second part of my talk\, I will discus
 s physical training strategies for neuromorphic systems [2\,3\,4].  Physic
 ally extracting the gradients required for training remains challenging as
  generic approaches only exist in certain cases. Equilibrium propagation (
 EP) is such a procedure that has been introduced and applied to classical 
 energy-based models which relax to an equilibrium. Here\, I will show a di
 rect connection between EP and Onsager reciprocity and exploit this to der
 ive a quantum version of EP [5]. This can be used to optimize loss functio
 ns that depend on the expectation values of observables of an arbitrary qu
 antum system. Specifically\, I will illustrate this new concept with super
 vised and unsupervised learning examples in which the input or the solvabl
 e task is of quantum mechanical nature\, e.g.\, the recognition of quantum
  many-body ground states\, quantum phase exploration\, sensing and phase b
 oundary exploration. We propose that in the future quantum EP may be used 
 to solve tasks such as quantum phase discovery with a quantum simulator ev
 en for Hamiltonians which are numerically hard to simulate or even partial
 ly unknown. Our scheme is relevant for a variety of quantum simulation pla
 tforms such as ion chains\, superconducting qubit arrays\, neutral atom Ry
 dberg tweezer arrays and strongly interacting atoms in optical lattices.  
 [1] C.C. Wanjura\, F. Marquardt. Nat Phys 20\, 1434–1440 (2024). [2] A. 
 Momeni\, B. Rahmani\, B. Scellier\, et al. Nature 645\, 53–61 (2025). [3
 ] Q. Wang\, C.C. Wanjura\, F. Marquardt. Neuromorph Comput Eng 4\, 034014 
 (2024). [4] N. Dal Cin\, F. Marquardt\, C.C. Wanjura. arXiv:2508.11750. [5
 ] C.C. Wanjura\, F. Marquardt. Nat Commun 16\, 6595 (2025).
DTSTAMP:20260512T124006Z
CREATED:20260320T063730Z
LAST-MODIFIED:20260330T053553Z
END:VEVENT
END:VCALENDAR