Co-Integration of an Analog, CMOS-Compatible Electro-Optical Conductive Metal-Oxide/HfO2 Memristor with Si Photonics
Open access
Date
2023-11-09Type
- Other Conference Item
ETH Bibliography
yes
Altmetrics
Abstract
Emulating biological learning rules through neuromorphic hardware is key to create energy-efficient computing systems for artificial intelligence. In this regard, neo-Hebbian three-factor learning rules are of particular interest as they describe the influence of global neuromodulators on synaptic plasticity and thus allow for biologically plausible realizations of reinforcement learning algorithms. Three-terminal memristors are ideal candidates to directly implement three-factor learning rules on-device, provided that the third terminal can modulate their conductance independently from the two conventional electrodes. Moreover, the emulation of synaptic plasticity also requires a continuous linear and symmetric conductance modulation. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000661334Publication status
publishedPublisher
ETH ZurichEvent
Organisational unit
03925 - Luisier, Mathieu / Luisier, Mathieu
Notes
Conference talk held on November 9, 2023.More
Show all metadata
ETH Bibliography
yes
Altmetrics