US 11,809,988 B2
Artificial intelligence system for classification of data based on contrastive learning
Anoop Cherian, Belmont, MA (US); and Aeron Shuchin, Medford, MA (US)
Assigned to Mitsubishi Electric Research Laboratories, Inc., Cambridge, MA (US)
Filed by Mitsubishi Electric Research Laboratories, Inc., Cambridge, MA (US)
Filed on Jun. 22, 2020, as Appl. No. 16/907,387.
Prior Publication US 2021/0397970 A1, Dec. 23, 2021
Int. Cl. G06N 3/08 (2023.01); G06N 3/088 (2023.01); G06F 17/18 (2006.01); G06F 18/21 (2023.01); G06F 18/24 (2023.01); G06N 3/045 (2023.01)
CPC G06N 3/08 (2013.01) [G06F 17/18 (2013.01); G06F 18/21 (2023.01); G06F 18/24 (2023.01); G06N 3/045 (2023.01); G06N 3/088 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An artificial intelligence (AI) system for classification of data including a processor configured to execute modules of the AI system, the modules comprising:
a feature extractor trained with machine learning to process input data to extract features of the input data for classification of the input data;
an adversarial noise generator trained with machine learning such that the adversarial noise generator is configured to generate noise data for distribution of features of the input data, wherein a misclassification rate of corrupted features that include extracted features corrupted with the generated noise data is caused by the training of the adversarial noise generator to be greater than a misclassification rate of the extracted features;
a compressor configured to compress the extracted features, wherein the compressed features are closer to the extracted features than to the corrupted features; and
a classifier trained with machine learning to classify the compressed features.