International Display Workshops General Incorporated Association

17:50 〜 18:10

[AMD5-3] An a-IGZO TFTs Based Neuron Circuit with Threshold Voltage Variation Compensation for Artificial Neural Network

*Yumin Yun1, Junhyeong Park1, Kyeong-Soo Kang1, Ji-Hwan Park1, Chanjin Park1, Soo-Yeon Lee1 (1. Seoul National University (Korea))

Neuron Circuit, Amorphous indium-gallium-zinc-oxide (a-IGZO), Threshold voltage compensation, Artificial Neural Network

https://doi.org/10.36463/idw.2023.0207

This work proposes an amorphous indium-gallium-zinc-oxide (a-IGZO) thin-film transistors (TFTs)-based neuron circuit for artificial neural networks. By using the threshold voltage (VTH) compensation method, the proposed circuit prevents the inaccurate output of the activation function caused by the threshold voltage (VTH) variation due to fabrication fluctuation.