An Augmented OxRAM Synapse for Spiking Neural Network (SNN) Circuits

Hassen Aziza (1, 2) , H. Bazzi (2, 3) , J. Postel-Pellerin (1, 2) , P. Canet (1, 2) , M. Moreau (2, 1) , A. Harb (3)

1 IM2NP - Institut des Matériaux, de Microélectronique et des Nanosciences de Provence
2 AMU - Aix Marseille Université
3 LIU - Lebanese International University

In this paper, the conductance modulation of OxRAM memristive devices is evaluated based on experimental data to reveal the memristor inherent analog synaptic behavior. Simulation results are presented to validate the use of OxRAMs as synapses at a circuit level in a spiking neural network context. In the proposed approach, the OxRAM synapse is augmented with a shift register associated with current compliance control transistors to provide an efficient monitoring of the OxRAM conductance.

Mots clés: Memristor; Neuromorphic circuits; OxRAM; Synapses; CMOS neurons; Neural networks

Domaines: Micro et nanotechnologies/Microélectronique; Intelligence artificielle [cs.AI]; Electronique

Hassen Aziza, H. Bazzi, J. Postel-Pellerin, P. Canet, M. Moreau, et al.. An Augmented OxRAM Synapse for Spiking Neural Network (SNN) Circuits. 2019 14th International Conference on Design & Technology of Integrated Systems In Nanoscale Era (DTIS), Apr 2019, Mykonos, France. ⟨10.1109/DTIS.2019.8735057⟩. ⟨hal-02306907⟩