US 11,816,186 B2
Architecture for dynamic ML model drift evaluation and visualization on a GUI
Philip A. Sallee, South Riding, VA (US); and Franklin Tanner, Ashburn, VA (US)
Assigned to Raytheon Company, Waltham, MA (US)
Filed by Raytheon Company, Waltham, MA (US)
Filed on Jul. 26, 2021, as Appl. No. 17/385,443.
Prior Publication US 2023/0025677 A1, Jan. 26, 2023
Int. Cl. G06F 18/21 (2023.01); G06F 3/0484 (2022.01); G06N 3/08 (2023.01)
CPC G06F 18/217 (2023.01) [G06F 3/0484 (2013.01); G06N 3/08 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for dynamic visualization of trained machine learning (ML) model performance on a graphical user interface (GUI), the method comprising:
receiving, from a user and by a user-adjustable artificial drift control of the GUI, data indicating a transformation to be applied to input samples;
applying the transformation to the input samples resulting in transformed input samples;
providing the transformed input samples to a trained ML model;
receiving data indicating the confidence and accuracy of the trained ML model with the transformed input samples as input, the confidence provided by the trained ML model and indicating an expected accuracy of a respective classifications by the trained ML model; and
dynamically displaying a concurrent plot of the accuracy and confidence to reflect the received data and as the received data are received.