US 11,669,730 B2
Recognition apparatus based on neural network and method of training neural network
Jun Haeng Lee, Hwaseong-si (KR)
Assigned to SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed by SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed on Oct. 30, 2019, as Appl. No. 16/668,712.
Application 16/668,712 is a continuation of application No. 15/195,302, filed on Jun. 28, 2016, granted, now 10,515,305.
Claims priority of application No. 10-2016-0009353 (KR), filed on Jan. 26, 2016.
Prior Publication US 2020/0065674 A1, Feb. 27, 2020
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/00 (2023.01); G06N 3/08 (2023.01); G06N 3/04 (2023.01); G06N 3/084 (2023.01); G06N 3/049 (2023.01); G06N 3/008 (2023.01); G06N 3/063 (2023.01); G06N 3/045 (2023.01)
CPC G06N 3/08 (2013.01) [G06N 3/008 (2013.01); G06N 3/04 (2013.01); G06N 3/045 (2023.01); G06N 3/049 (2013.01); G06N 3/063 (2013.01); G06N 3/084 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A recognition apparatus comprising:
a memory configured to store a neural network comprising a previous layer of neurons, and a current layer of neurons that are activated based on first synaptic signals and second synaptic signals, the first synaptic signals being input from the previous layer, and the second synaptic signals being input from the current layer; and
a processor configured to generate a recognition result based on the neural network,
wherein the neurons of the current layer in the neural network are classified into at least one group, and an activation neuron among neurons included in the same group generates a second synaptic signal to inhibit other neurons in the same group, and
wherein the processor is further configured to change a membrane potential time constant value of neurons included in the neural network based on a comparison between a magnitude of input signals that are input to the neural network and each of an upper reference value and a lower reference value.