Novel computational element based on VO2 oscillators: From material, beyond material

NISE Seminar

  • Date: Nov 21, 2024
  • Time: 10:30 AM - 11:30 AM (Local Time Germany)
  • Speaker: Guanmin Li
  • Location: Max-Planck-Institut für Mikrostrukturphysik, Weinberg 2, 06120 Halle (Saale)
  • Room: Lecture Hall, B.1.11
Novel computational element based on VO<sub>2</sub> oscillators: From material, beyond material

Computational technologies based on coupled oscillators are of great interest for energy efficient computing [1, 2]. A key to the development of such technologies is the tunable control of the interaction strength between the oscillators. Thus far, such coupled oscillators have been accomplished by additional external electronic components [3-5].

Here [6] we show that the synchronization of closely spaced vanadium dioxide (VO2) oscillators can be controlled via a thermal triggering element that itself is formed from VO2. This tuning process allows the control of the amplitude, frequency of the coupled VO2 oscillators. 3 Boolean logic gates AND, NAND and NOR are realized by these oscillators. Biological plausibility and functionality of two spiking neuron models (Hodgkin-Huxley model and leaky-integrate-and-fire model) directly implemented by VO2 oscillators are demonstrated. Simulation of network based on 150 VO2 LIF neurons shows 90% accuracy for recognizing hand-written digits from MNIST data set. Our findings demonstrate that the networks of thermally coupled oscillators can be readily applicable to novel bio-inspired computational schemes.

References

  1. [1] M. Romera, P. Talatchian, S. Tsunegi et al, Nature (2018) 563, 230-234.

  2. [2] A. Raychowdhury, A. Parihar et al, Proc IEEE (2019) 107, 73-89.

  3. [3] E. Corti, C. Antonio et al, Front. Neurosci. (2021) 15, 628254.

  4. [4] R. Yuan, Q. Duan, P.J. Tiw et al. Nat. Commun. (2022) 13, 3973.

  5. [5] J. Núñez, A. María J et al, Front. Neurosci. (2021) 15, 655823.

  6. [6] G. Li et al, Nat Commun (2024) 15, 5820.

Go to Editor View