US 12,242,530 B2
Hyper-personalized prompt based content generation
Yuan He, San Jose, CA (US); Anupam Anurag Tripathi, Chicago, IL (US); Anwitha Paruchuri, San Jose, CA (US); Sukryool Kang, Dublin, CA (US); Andrew Francis Hickl, Sammamish, WA (US); Sujeong Cha, Long Island City, NY (US); Surya Raghavendra Vadlamani, Newtown, PA (US); and Peter Royer Smith, Jr., Sharon, CT (US)
Assigned to Accenture Global Solutions Limited, Dublin (IE)
Filed by Accenture Global Solutions Limited, Dublin (IE)
Filed on Apr. 28, 2023, as Appl. No. 18/140,812.
Prior Publication US 2024/0362265 A1, Oct. 31, 2024
Int. Cl. G06F 16/535 (2019.01); G06F 16/538 (2019.01); G06F 16/56 (2019.01); G06N 3/08 (2023.01)
CPC G06F 16/535 (2019.01) [G06F 16/538 (2019.01); G06F 16/56 (2019.01); G06N 3/08 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for generating a personalized image, comprising:
generating a personalized text prompt by processing an input embedding using a transformer model followed by a first fully connected neural network, wherein the input embedding comprises a multi-dimensional embedding vector associated with a user profile and a plurality of user items;
generating a scored label set identifying a user's preferences by processing a set of attributes for the plurality of user items using a second fully connected neural network that is different from the first fully connected neural network, wherein:
the scored label set includes a plurality of target labels and a score assigned to each of the plurality of target labels, and
the score of a target label of the plurality of target labels characterizes the user's preferences corresponding to the target label; and
generating the personalized image by processing the personalized text prompt and the scored label set using a diffusion model.