US 11,676,036 B2
Apparatus for knowledge based evolutionary learning in AI systems
Nagarjun Pogakula Surya, Bangalore (IN); Gomathi Sankar, Bangalore (IN); Fuk Ho Pius Ng, Portland, OR (US); and Satish Padmanabhan, Sunnyvale, CA (US)
Assigned to DIMAAG-AI, Inc., Fremont, CA (US)
Filed by DIMAAG-AI, Inc., Fremont, CA (US)
Filed on May 21, 2020, as Appl. No. 16/880,346.
Prior Publication US 2021/0365793 A1, Nov. 25, 2021
Int. Cl. G06N 3/086 (2023.01); G06N 3/045 (2023.01)
CPC G06N 3/086 (2013.01) [G06N 3/045 (2023.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
determining via a processor an initial trained neural network based on initial training data that includes a plurality of training data observations, each of the training data observations including a respective plurality of initial training data feature values and a respective initial training data observed outcome value;
determining via the processor a plurality of initial training data predicted outcome values by applying the initial trained neural network to the initial training data feature values;
determining via the processor an updated trained neural network based on updated training data, the updated training data including a first portion including incremental training data not included in the initial training data, the updated training data including a second portion including the plurality of initial training data feature values and the plurality of initial training predicted outcome values;
determining via the processor one or more live data predicted outcome values by applying the updated trained neural network to live data; and
storing the live data predicted outcome values on a storage device.