US 12,039,428 B2
Multi-model based target engagement sequence generator
Jere Armas Michael Helenius, Cupertino, CA (US); Nandan Gautam Thor, Mountain View, CA (US); Erik Michael Bower, San Francisco, CA (US); and René Bonvanie, Foster City, CA (US)
Assigned to Palo Alto Networks, Inc., Santa Clara, CA (US)
Filed by Palo Alto Networks, Inc., Santa Clara, CA (US)
Filed on Sep. 15, 2022, as Appl. No. 17/932,365.
Application 17/932,365 is a continuation of application No. 16/371,098, filed on Mar. 31, 2019, granted, now 11,494,610.
Prior Publication US 2023/0011066 A1, Jan. 12, 2023
Int. Cl. G06N 3/044 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06Q 30/0202 (2023.01)
CPC G06N 3/044 (2023.01) [G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06Q 30/0202 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
determining a plurality of sequences of organizational classifications for a set of opportunity characteristics;
retrieving historical time series data corresponding to the set of opportunity characteristics, wherein the historical time series data comprises data for a plurality of individuals and a plurality of opportunities;
converting the historical time series data into first training data for a statistical model, wherein the first training data comprises a plurality of sale values for each of the plurality of sequences of organizational classifications indicated in the historical time series data;
converting the historical time series data into second training data for a recurrent neural network, wherein the second training data comprises the plurality of sequences of organizational classifications and indications of success for each of the plurality of sequences of organizational classifications on opportunities corresponding to the set of opportunity characteristics;
training the recurrent neural network on the second training data to output a sequence of organizational classifications that increases likelihood of conversion;
training the statistical model on the first training data to output relative importance values for organizational classifications in the plurality of sequences of organizational classifications for the set of opportunity characteristics; and
at least one of inserting in memory and updating in memory the trained recurrent neural network and the trained statistical model for generation of sequences of individuals to engage that increase success of opportunities satisfying the set of opportunity characteristics.