US 12,450,478 B2
Dynamic data-dependent neural network processing systems and methods
Mark Alan Lovell, Lucas, TX (US); and Robert Michael Muchsel, Addison, TX (US)
Assigned to Maxim Integrated Products, Inc., San Jose, CA (US)
Filed by Maxim Integrated Products, Inc., San Jose, CA (US)
Filed on Sep. 10, 2021, as Appl. No. 17/472,074.
Prior Publication US 2023/0077454 A1, Mar. 16, 2023
Int. Cl. G06N 3/08 (2023.01); G06N 3/049 (2023.01); G06V 10/75 (2022.01)
CPC G06N 3/08 (2013.01) [G06N 3/049 (2013.01); G06V 10/751 (2022.01)] 18 Claims
OG exemplary drawing
 
1. A dynamic data-dependent neural network processing method comprising:
at a controller, receiving input data that is to be processed in a first layer in a sequence of processing layers of a neural network by using a first set of parameters, the first layer to be processed by a first device;
analyzing the input data in the controller to make a determination regarding at least one of:
whether to obtain modified input data that would increase computational efficiency of at least one computational resource;
whether to process at least one of the input data or the modified input data in a second layer would increase computational efficiency of the at least one computational resource; or
whether to apply a second set of parameters that is different from the first set of parameters to at least one of the input data or the modified input data would increase computational efficiency of the at least one computational resource;
in response to the determination, modifying the sequence of processing layers of the neural network to obtain a modified sequence that uses each of the processing layers; and
processing at least one of the input data or the modified input data according to the modified sequence to cause the neural network to behave differently depending on the input data to increase computational efficiency of the at least one computational resource.