	US 11,954,579 B2
	Synaptic weight training method, target identification method, electronic device and medium
Zhenzhi Wu, Beijing (CN); Xin Ma, Beijing (CN); Qune Kong, Beijing (CN); and Yaolong Zhu, Beijing (CN)
Assigned to LYNXI TECHNOLOGIES CO., LTD., Beijing (CN)
Appl. No. 18/265,715
Filed by LYNXI TECHNOLOGIES CO., LTD., Beijing (CN)
PCT Filed May 31, 2022, PCT No. PCT/CN2022/096281 § 371(c)(1), (2) Date Jun. 7, 2023, PCT Pub. No. WO2022/253229, PCT Pub. Date Dec. 8, 2022.
Claims priority of application No. 202110624089.9 (CN), filed on Jun. 4, 2021; application No. 202110625816.3 (CN), filed on Jun. 4, 2021; and application No. 202110625880.1 (CN), filed on Jun. 4, 2021.
Prior Publication US 2023/0394288 A1, Dec. 7, 2023
Int. Cl. G06N 3/049 (2023.01); G06N 3/084 (2023.01)

CPC G06N 3/049 (2013.01) [G06N 3/084 (2013.01)]

19 Claims

1. A synaptic weight training method, comprising:

inputting spike signals corresponding to training samples into a spiking neural network to be trained;

learning a target synaptic weight in the spiking neural network with a back propagation rule to obtain a first branch weight; and learning the target synaptic weight with a synaptic plasticity rule to obtain a second branch weight; and

updating the target synaptic weight according to the first branch weight and the second branch weight,

wherein after learning the target synaptic weight in the spiking neural network with the back propagation rule to obtain the first branch weight, the method further comprises:

writing the first branch weight into a first branch weight buffer area;

after learning the target synaptic weight with the synaptic plasticity rule to obtain the second branch weight, the method further comprises:

writing the second branch weight into a second branch weight buffer area.