US 12,437,510 B2
Utilizing machine learning models to identify implicit bias
Anutosh Maitra, Bangalore (IN); Milind Savagaonkar, Bangalore (IN); Sowmya Rasipuram, Bangalore (IN); Roshni Ramesh Ramnani, Bangalore (IN); Karthik L Nagar, Shivamogga (IN); and Aditya Mohan Singh, Halanayakanahalli (IN)
Assigned to Accenture Global Solutions Limited, Dublin (IE)
Filed by Accenture Global Solutions Limited, Dublin (IE)
Filed on Dec. 20, 2022, as Appl. No. 18/068,945.
Prior Publication US 2024/0203094 A1, Jun. 20, 2024
Int. Cl. G06V 10/764 (2022.01); G06V 10/70 (2022.01); G06V 10/774 (2022.01); G10L 15/26 (2006.01)
CPC G06V 10/764 (2022.01) [G06V 10/768 (2022.01); G06V 10/774 (2022.01); G10L 15/26 (2013.01)] 18 Claims
OG exemplary drawing
 
7. A device, comprising:
one or more memories; and
one or more processors, coupled to the one or more memories, configured to:
receive multimodal content that includes two or more of text data, image data, or video data;
process the two or more of the text data, the image data, or the video data of the multimodal content, with a binary classifier model, to determine whether the multimodal content is biased or unbiased;
process the two or more of the text data, the image data, or the video data of the multimodal content, with a multilabel classifier model and based on the multimodal content being biased, to determine an implicit bias in the multimodal content; and
perform one or more actions based on the implicit bias; and
utilize a plurality of image models to identify image tag line data, image caption data, and image attribute data in the image data,
wherein processing the two or more of the text data, the image data, or the video data, with the binary classifier model comprises:
processing the image tag line data, the image caption data, and the image attribute data, with the binary classifier model, to determine whether the image data is biased or unbiased,
wherein processing the two or more of the text data, the image data, or the video data with the multilabel classifier model comprises:
processing the image tag line data, the image caption data, and the image attribute data, with the multilabel classifier model and based on the image data being biased, to determine an implicit image bias.