Simulating anti-Hebbian Spike Time Dependent Plasticity in bottom-gated polymer-wrapped carbon nanotube synaptic transistors

Flohil, René (2021) Simulating anti-Hebbian Spike Time Dependent Plasticity in bottom-gated polymer-wrapped carbon nanotube synaptic transistors. Master's Thesis / Essay, Human-Machine Communication.

Preview	Text Master_Thesis_RFlohil_v020921.pdf Download (4MB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (125kB)

Abstract

In a recent paper by Talsma et al. (2020) anti-Hebbian Spike Time Dependent Plasticity (STDP) results were produced in a semiconducting single-walled carbon nanotube (s-SWCNT) inked bottom gated field-effect transistor (FET) by pulsing the device with pulse-pairs with varying delays. An attempt is made to reproduce these results using data on the plasticity of the device, but this proved ineffective. Therefore two new experiments were performed to mea- sure the plasticity of a similar device when pulsed with varying pulse lengths and varying gate voltages. Using two Generalized Additive Models (GAMs) to describe the non-linear relation between pulsing and conductance for positive and negative pulsing it is possible to produce source-drain current values. The simulation is able to give insight into how the characteristics of the device affect weight change using the source-gate bias as the main factor in conductance change by producing a weight change graph over varying delays between the pre- and postpulse. The anti-Hebbian results are the result of a misassignment of the pre- and post-synaptic labels to the source-drain and gate terminals, in actuality the synaptic transistor produces Hebbian STDP results. The simulation produces STDP-like results for negative delays due to the polarity switch of the bias pulse when the delay becomes negative. This would make the device unable to perform proper STDP and thus less suitable for integration into artificial neural networks.

Item Type:	Thesis (Master's Thesis / Essay)
Supervisor name:	Taatgen, N.A. and Borst, J.P.
Degree programme:	Human-Machine Communication
Thesis type:	Master's Thesis / Essay
Language:	English
Date Deposited:	21 Sep 2021 09:25
Last Modified:	21 Sep 2021 09:25
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/26111

Actions (login required)

View Item