US 11,816,566 B2
Joint learning from explicit and inferred labels
Subhabrata Mukherjee, Seattle, WA (US); Guoqing Zheng, Redmond, WA (US); Ahmed Awadalla, Redmond, WA (US); Milad Shokouhi, Seattle, WA (US); Susan Theresa Dumais, Kirkland, WA (US); and Kai Shu, Mesa, AZ (US)
Assigned to Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on May 18, 2020, as Appl. No. 16/876,931.
Prior Publication US 2021/0357747 A1, Nov. 18, 2021
Int. Cl. G06V 10/82 (2022.01); G06N 3/08 (2023.01); G06N 3/04 (2023.01)
CPC G06N 3/08 (2013.01) [G06N 3/04 (2013.01); G06V 10/82 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A method performed on a computing device, the method comprising:
providing a machine learning model having a first classification layer, a second classification layer, and an encoder that feeds into the first classification layer and the second classification layer;
obtaining first training examples having explicit labels and second training examples having inferred labels, wherein the inferred labels are based at least on actions associated with the second training examples;
training the machine learning model using the first training examples and the second training examples using a training objective, wherein the training objective considers first training loss of the first classification layer for the explicit labels and second training loss of the second classification layer for the inferred labels; and
outputting a trained machine learning model having at least the encoder and the first classification layer,
wherein the encoder is configured to map the first training examples and the second training examples into a shared vector space.