Trainable neuromorphic spintronic hardware Via analog finite-difference gradient methods
Catarina Pereira, Alex Jenkins, Eleonora Raimondo, Mario Carpentieri, Ensieh Iranmehr, Luana Benetti, Subhajit Roy, Ricardo Ferreira, Joao Ventura, Giovanni Finocchio, Davi Rodrigues · 2026
Read Trainable neuromorphic spintronic hardware Via analog finite-difference gradient methods by Catarina Pereira, Alex Jenkins, Eleonora Raimondo, Mario Carpentieri, Ensieh Iranmehr, Luana Benetti, Subhajit Roy, Ricardo Ferreira, Joao Ventura, Giovanni Finocchio, Davi Rodrigues (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13385967.txt.
Description
Catarina Pereira's team built trainable spin hardware that learns on-device, in a 2026 study. Spin nano-neurons promise energy-efficient neural hardware, but training relied on oversimplified device models sensitive to variation. The team showed magnetic tunnel junctions can generate tunable complex nonlinear responses, then implemented analog finite-difference gradient generation on-chip.
Device-in-the-loop training in a magnetic tunnel junction network with one and two hidden layers reached 93.3 percent classification accuracy despite marked device variation. Gradients from the analog neurons closely matched numerical values. The design removes the training bottleneck for spin neural networks, bringing hardware learning closer to practice.
Catalog record
| CREATOR | Catarina Pereira, Alex Jenkins, Eleonora Raimondo, Mario Carpentieri, Ensieh Iranmehr, Luana Benetti, Subhajit Roy, Ricardo Ferreira, Joao Ventura, Giovanni Finocchio, Davi Rodrigues |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |