US 11,790,044 B2
Detecting anomalous events in a discriminator of an embedded device
Javier Elenes, Austin, TX (US); and Antonio Torrini, Austin, TX (US)
Assigned to Silicon Laboratories Inc., Austin, TX (US)
Filed by Silicon Laboratories Inc., Austin, TX (US)
Filed on Jun. 23, 2022, as Appl. No. 17/847,408.
Application 17/847,408 is a division of application No. 17/089,193, filed on Nov. 4, 2020, granted, now 11,403,496.
Prior Publication US 2022/0335260 A1, Oct. 20, 2022
Int. Cl. G06F 18/21 (2023.01); G06N 3/088 (2023.01); G06F 18/25 (2023.01); G06F 18/20 (2023.01); G06F 18/214 (2023.01)
CPC G06F 18/2193 (2023.01) [G06F 18/2148 (2023.01); G06F 18/251 (2023.01); G06F 18/285 (2023.01); G06N 3/088 (2013.01)] 17 Claims
OG exemplary drawing
 
1. An apparatus comprising:
a sensor to sense real world information;
a digitizer coupled to the sensor to digitize the real world information into digitized information;
a signal processor coupled to the digitizer to process the digitized information into an image;
a discriminator comprising a binary classifier coupled to the signal processor, the discriminator to determine, based at least in part on the image, whether the real world information comprises an anomaly, wherein the discriminator is trained via a generative adversarial network with samples of a positive condition comprising the anomaly, the samples of the anomaly comprising synthetic samples generated by a generator of the generative adversarial network and real samples of the anomaly, wherein the binary classifier is to output a probability value, wherein when the probability value exceeds a threshold the anomaly is detected and when the probability value does not exceed the threshold no anomaly is detected; and
a controller coupled to the discriminator, wherein in response to an indication of the anomaly from the discriminator, the controller is to trigger an action.
 
12. A system comprising:
a sensor to sense real world information;
a digitizer coupled to the sensor to digitize the real world information into digitized information;
a signal processor coupled to the digitizer to process the digitized information into an image;
a discriminator coupled to the signal processor, the discriminator to determine, based at least in part on the image, whether the real world information comprises an anomaly, wherein the discriminator is trained via a generative adversarial network having a generator that provides a plurality of random images to the discriminator, the plurality of random images identified to the discriminator as real images;
an actuator to perform an action; and
a controller coupled to the discriminator, wherein in response to an indication of the anomaly from the discriminator, the controller is to cause the actuator to perform the action.
 
15. A wireless device comprising:
a sensor to sense real world information;
a digitizer coupled to the sensor to digitize the real world information into digitized information;
a signal processor coupled to the digitizer to process the digitized information into an image;
a discriminator coupled to the signal processor, the discriminator to determine, based at least in part on the image, whether the real world information comprises an anomaly, wherein the discriminator is trained via a generative adversarial network located externally to the wireless device, the wireless device not including the generative adversarial network, the generative adversarial network to provide a plurality of random images to the discriminator, the plurality of random images identified to the discriminator as real images; and
a wireless circuit coupled to the signal processor, wherein the wireless circuit is to communicate information with a remote device.