US 12,217,155 B2
Deep neural network architecture using piecewise linear approximation
Kamlesh Pillai, Bangalore (IN); Gurpreet S. Kalsi, Bangalore (IN); and Amit Mishra, Bangalore (IN)
Assigned to Intel Corporation, Santa Clara, CA (US)
Filed by Intel Corporation, Santa Clara, CA (US)
Filed on Aug. 24, 2023, as Appl. No. 18/455,026.
Application 18/455,026 is a continuation of application No. 16/023,441, filed on Jun. 29, 2018, granted, now 11,775,805.
Prior Publication US 2024/0020518 A1, Jan. 18, 2024
Int. Cl. G06N 3/048 (2023.01); G06F 7/499 (2006.01); G06F 7/556 (2006.01); G06F 17/11 (2006.01); G06F 17/17 (2006.01); G06N 3/044 (2023.01); G06N 3/045 (2023.01); G06N 3/063 (2023.01); G06N 3/084 (2023.01)
CPC G06N 3/048 (2023.01) [G06F 7/49957 (2013.01); G06F 7/556 (2013.01); G06F 17/11 (2013.01); G06F 17/17 (2013.01); G06N 3/045 (2023.01); G06N 3/063 (2013.01); G06N 3/084 (2013.01); G06N 3/044 (2023.01)] 20 Claims
OG exemplary drawing
 
1. A method for implementing a deep neural network, the method comprising:
receiving an input vector to be processed in the deep neural network, the input vector comprising one or more floating-point input elements, the deep neural network comprising one or more hidden layers;
converting the one or more floating-point input elements into one or more fixed-point input elements;
inputting the one or more fixed-point input elements into the one or more hidden layers;
computing, by the one or more hidden layers, one or more fixed-point output elements by:
converting one or more floating-point weights of the one or more hidden layers into one or more fixed-point weights, and
performing, in the one or more hidden layers, a convolution using the one or more fixed-point weights and the one or more fixed-point input elements;
converting the one or more fixed-point output elements into one or more floating-point output elements; and
generating an output of the deep neural network using the one or more floating-point output elements.