001025209 001__ 1025209
001025209 005__ 20250203103227.0
001025209 0247_ $$2doi$$a10.29363/nanoge.neuronics.2024.027
001025209 037__ $$aFZJ-2024-02777
001025209 1001_ $$0P:(DE-Juel1)188273$$aNeftci, Emre$$b0$$ufzj
001025209 1112_ $$aNeuronics Conference$$cValència$$d2024-02-21 - 2024-02-23$$wSpain
001025209 245__ $$aPre-training and Meta-learning for Memristor Crossbar Arrays
001025209 260__ $$c2023
001025209 3367_ $$033$$2EndNote$$aConference Paper
001025209 3367_ $$2DataCite$$aOther
001025209 3367_ $$2BibTeX$$aINPROCEEDINGS
001025209 3367_ $$2DRIVER$$aconferenceObject
001025209 3367_ $$2ORCID$$aLECTURE_SPEECH
001025209 3367_ $$0PUB:(DE-HGF)6$$2PUB:(DE-HGF)$$aConference Presentation$$bconf$$mconf$$s1714556807_3947$$xInvited
001025209 520__ $$aMemristive crossbar arrays show promise as non-von Neumann computing technologies, bringing sophisticated neural network processing to the edge and facilitating real-world online learning. However, their deployment for real-world learning problems faces challenges such as non-linearities in conductance updates, variations during operation, fabrication mismatch, conductance drift, and the realities of gradient descent training.This talk will present methods to pre-train neural networks to be largely insensitive to these non-idealities during learning tasks. These methods rely on a phenomenological model of the device, obtainable experimentally, and bi-level optimization. We showcase this effect through meta-learning and a differentiable model of conductance updates on few-shot learning tasks. Since pre-training is a necessary procedure for any online learning scenario at the edge, our results may pave the way for real-world applications of memristive devices without significant adaptation overhead.Furthermore, by considering the programming of memristive devices as a learning problem in its own right, we demonstrate that the developed methods can accelerate existing write-verify techniques.
001025209 536__ $$0G:(DE-HGF)POF4-5234$$a5234 - Emerging NC Architectures (POF4-523)$$cPOF4-523$$fPOF IV$$x0
001025209 588__ $$aDataset connected to CrossRef Conference
001025209 773__ $$a10.29363/nanoge.neuronics.2024.027
001025209 8564_ $$uhttps://www.nanoge.org/proceedings/Neuronics/656741f369ece060987dcf51
001025209 909CO $$ooai:juser.fz-juelich.de:1025209$$pVDB
001025209 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)188273$$aForschungszentrum Jülich$$b0$$kFZJ
001025209 9131_ $$0G:(DE-HGF)POF4-523$$1G:(DE-HGF)POF4-520$$2G:(DE-HGF)POF4-500$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$9G:(DE-HGF)POF4-5234$$aDE-HGF$$bKey Technologies$$lNatural, Artificial and Cognitive Information Processing$$vNeuromorphic Computing and Network Dynamics$$x0
001025209 9141_ $$y2024
001025209 9201_ $$0I:(DE-Juel1)PGI-15-20210701$$kPGI-15$$lNeuromorphic Software Eco System$$x0
001025209 980__ $$aconf
001025209 980__ $$aVDB
001025209 980__ $$aI:(DE-Juel1)PGI-15-20210701
001025209 980__ $$aUNRESTRICTED