US20220138499A1 - MAXIMIZING THE OPERATIONAL RANGE FOR TRAINING PARAMETERS WHILE SELECTING TRAINING VECTORS FOR A MACHINE-LEARNING MODEL