Matches in SemOpenAlex for { <https://semopenalex.org/work/W4310285122> ?p ?o ?g. }
Showing items 1 to 64 of
64
with 100 items per page.
- W4310285122 abstract "This project evaluated the use of emerging spintronic memory devices for robust and efficient variational inference schemes. Variational inference (VI) schemes, which constrain the distribution for each weight to be a Gaussian distribution with a mean and standard deviation, are a tractable method for calculating posterior distributions of weights in a Bayesian neural network such that this neural network can also be trained using the powerful backpropagation algorithm. Our project focuses on domain-wall magnetic tunnel junctions (DW-MTJs), a powerful multi-functional spintronic synapse design that can achieve low power switching while also opening the pathway towards repeatable, analog operation using fabricated notches . Our initial efforts to employ DW-MTJs as an all-in-one stochastic synapse with both a mean and standard deviation didn?t end up meeting the quality metrics for hardware-friendly VI. In the future, new device stacks and methods for expressive anisotropy modification may make this idea still possible. However , as a fall back that immediately satisfies our requirements, we invented and detailed how the combination of a DW-MTJ synapse encoding the mean and a probabilistic Bayes-MTJ device , programmed via a ferroelectric or ionically modifiable layer, can robustly and expressively implement VI. This design includes a physics- informed small circuit model, that was scaled up to perform and demonstrate rigorous uncertainty quantification applications, up to and including small convolutional networks on a grayscale image classification task, and larger (Residual) networks implementing multi-channel image classification. Lastly, as these results and ideas all depend upon the idea of an inference application where weights (spintronic memory states) remain non-volatile, the retention of these synapses for the notched case was further interrogated. These investigations revealed and emphasized the importance of both notch geometry and anisotropy modification in order to further enhance the endurance of written spintronic states. In the near future, these results will be mapped to effective predictions for room temperature and elevated operation DW-MTJ memory retention, and experimentally verified when devices become available." @default.
- W4310285122 created "2022-11-30" @default.
- W4310285122 creator A5001046383 @default.
- W4310285122 creator A5003734782 @default.
- W4310285122 creator A5011371394 @default.
- W4310285122 creator A5013365606 @default.
- W4310285122 creator A5042444120 @default.
- W4310285122 creator A5052112271 @default.
- W4310285122 creator A5076971323 @default.
- W4310285122 creator A5086398411 @default.
- W4310285122 date "2022-09-01" @default.
- W4310285122 modified "2023-10-16" @default.
- W4310285122 title "Probabilistic Nanomagnetic Memories for Uncertain and Robust Machine Learning." @default.
- W4310285122 doi "https://doi.org/10.2172/1891190" @default.
- W4310285122 hasPublicationYear "2022" @default.
- W4310285122 type Work @default.
- W4310285122 citedByCount "0" @default.
- W4310285122 crossrefType "report" @default.
- W4310285122 hasAuthorship W4310285122A5001046383 @default.
- W4310285122 hasAuthorship W4310285122A5003734782 @default.
- W4310285122 hasAuthorship W4310285122A5011371394 @default.
- W4310285122 hasAuthorship W4310285122A5013365606 @default.
- W4310285122 hasAuthorship W4310285122A5042444120 @default.
- W4310285122 hasAuthorship W4310285122A5052112271 @default.
- W4310285122 hasAuthorship W4310285122A5076971323 @default.
- W4310285122 hasAuthorship W4310285122A5086398411 @default.
- W4310285122 hasBestOaLocation W43102851222 @default.
- W4310285122 hasConcept C11413529 @default.
- W4310285122 hasConcept C121332964 @default.
- W4310285122 hasConcept C151927369 @default.
- W4310285122 hasConcept C154945302 @default.
- W4310285122 hasConcept C163716315 @default.
- W4310285122 hasConcept C41008148 @default.
- W4310285122 hasConcept C49937458 @default.
- W4310285122 hasConcept C50644808 @default.
- W4310285122 hasConcept C62520636 @default.
- W4310285122 hasConcept C81363708 @default.
- W4310285122 hasConceptScore W4310285122C11413529 @default.
- W4310285122 hasConceptScore W4310285122C121332964 @default.
- W4310285122 hasConceptScore W4310285122C151927369 @default.
- W4310285122 hasConceptScore W4310285122C154945302 @default.
- W4310285122 hasConceptScore W4310285122C163716315 @default.
- W4310285122 hasConceptScore W4310285122C41008148 @default.
- W4310285122 hasConceptScore W4310285122C49937458 @default.
- W4310285122 hasConceptScore W4310285122C50644808 @default.
- W4310285122 hasConceptScore W4310285122C62520636 @default.
- W4310285122 hasConceptScore W4310285122C81363708 @default.
- W4310285122 hasLocation W43102851221 @default.
- W4310285122 hasLocation W43102851222 @default.
- W4310285122 hasOpenAccess W4310285122 @default.
- W4310285122 hasPrimaryLocation W43102851221 @default.
- W4310285122 hasRelatedWork W2521062615 @default.
- W4310285122 hasRelatedWork W2735477435 @default.
- W4310285122 hasRelatedWork W2807436399 @default.
- W4310285122 hasRelatedWork W3016958897 @default.
- W4310285122 hasRelatedWork W3045739591 @default.
- W4310285122 hasRelatedWork W3181746755 @default.
- W4310285122 hasRelatedWork W4283379348 @default.
- W4310285122 hasRelatedWork W4290792893 @default.
- W4310285122 hasRelatedWork W4312417841 @default.
- W4310285122 hasRelatedWork W4386750284 @default.
- W4310285122 isParatext "false" @default.
- W4310285122 isRetracted "false" @default.
- W4310285122 workType "report" @default.