New patent application published for: FALSE ALARM CLASSIFICATION

FALSE ALARM CLASSIFICATION

Claims

1. A method, comprising: receiving an alarm triggered by a component within an environment monitored by a security system; generating one or more event outputs for the environment comprising at least three of: generating a trigger-driven output by processing trigger-driven data using a trigger-driven fault detection model, generating a laser-based output by processing laser-based data using a laser fault detection model, generating an image-based output by processing image-based data using an image fault detection model, or generating a radio frequency (RF)-based output by processing RF-based data using a RF fault detection model; comparing the one or more event outputs, comprising at least three of the trigger-driven output, the laser-based output, the image-based output, and the RF-based output, with a rule-based table; and generating a likelihood of the alarm being a false alarm based on the comparison.
2. The method of claim 1, further comprising trigging a security response based on the likelihood of the alarm being a false alarm.
3. The method of claim 1, wherein the component is selected from one or more trigger-driven sensors, one or more laser-based sensors, one or more cameras, and one or more RF-based sensors.
4. The method of claim 1, wherein each of the trigger-driven output, the laser-based output, the image-based output, and the RF-based output is selected from: an indicator representing the alarm is faulty and a respective trigger-driven sensor, laser-based sensor, camera, and RF-based sensor is malfunctioning, an indicator representing no alarm and the respective trigger-driven sensor, laser-based sensor, camera, or RF-based sensor is properly functioning, or an indicator representing the alarm is valid and the respective trigger-driven sensor, laser-based sensor, camera, or RF-based sensor is properly functioning.
5. The method of claim 1, wherein the rule-based table defines a plurality of alarm classifications based on a combination of the trigger-driven output, the laser-based output, the image-based output, and the RF-based output.
6. The method of claim 1, wherein the rule-based table defines a plurality of alarm classifications based on a combination of the trigger-driven output, the laser-based output, and the image-based output.
7. The method of claim 1, wherein the rule-based table defines a plurality of alarm classifications based on a combination of the trigger-driven output, the laser-based output, and the RF-based output.
8. The method of claim 1, wherein the trigger-driven fault detection model is trained by: collecting historical trigger-driven sensor data from the environment; allocating a label to the historical trigger-driven sensor data based on an output of a historical alarm, wherein the label is selected from: an indicator representing the historical alarm is faulty and a respective trigger-driven sensor is malfunctioning, an indicator representing no alarm and the respective trigger-driven sensor is properly functioning, or an indicator representing the historical alarm is valid and the respective trigger-driven sensor is properly functioning; and training the trigger-driven fault detection model by using the historical trigger-driven sensor data as an input and the label as a target output, wherein the trigger-driven fault detection model learns to correlate the input to the target output.
9. The method of claim 1, wherein the laser fault detection model is trained by: collecting historical laser-based data from the environment; allocating a label to the historical laser-based data based on an output of a historical alarm, wherein the label is selected from: an indicator representing the historical alarm is faulty and a respective laser-based sensor is malfunctioning, an indicator representing no alarm and the respective laser-based sensor is properly functioning, or an indicator representing the historical is valid and the respective laser-based sensor is properly functioning; and training the laser fault detection model by using the historical laser-based data as an input and the label as a target output, wherein the laser fault detection model learns to correlate the input to the target output.
10. The method of claim 1, wherein the image fault detection model is trained by: collecting historical image data from the environment; allocating a label to the historical image data based on an output of a historical alarm, wherein the label is selected from: an indicator representing the historical alarm is faulty and a respective camera is malfunctioning, an indicator representing no alarm and the respective camera is properly functioning, or an indicator representing the historical alarm is valid and the respective camera is properly functioning; and training the image fault detection model by using the historical image data as an input and the label as a target output, wherein the image fault detection model learns to correlate the input to the target output.
11. The method of claim 1, wherein the RF fault detection model is trained by: collecting historical RF-based data from the environment; allocating a label to the historical RF-based data based on an output of a historical alarm, wherein the label is selected from: an indicator representing the historical alarm is faulty and a respective RF-based sensor is malfunctioning, an indicator representing no alarm and the respective RF-based sensor is properly functioning, or an indicator representing the historical alarm is valid and the respective RF-based sensor is properly functioning; and training the RF fault detection model by using the historical RF-based data as an input and the label as a target output, wherein the RF fault detection model learns to correlate the input to the target output.
12. A system, comprising: one or more memories collectively containing one or more programs; and one or more processors, wherein the one or more processors are configured to, individually or collectively, perform an operation comprising: receiving an alarm triggered by a component within an environment monitored by a security system; generating one or more event outputs for the environment comprising at least three of: generating a trigger-driven output by processing trigger-driven data using an trigger-driven fault detection model, generating a laser-based output by processing laser-based data using a laser fault detection model, generating an image-based output by processing image-based data using an image fault detection model, or generating a radio frequency (RF)-based output by processing RF-based data using a RF fault detection model; comparing the one or more event outputs, comprising at least three of the trigger-driven output, the laser-based output, the image-based output, and the RF-based output, with a rule-based table; and generating a likelihood of the alarm being a false alarm based on the comparison.
13. The system of claim 12, wherein the operation further comprises trigging a security response based on the likelihood of the alarm being a false alarm.
14. The system of claim 12, wherein the component is selected from one or more trigger-driven sensors, one or more laser-based sensors, one or more cameras, and one or more RF-based sensors.
15. The system of claim 12, wherein each of the trigger-driven output, the laser-based output, the image-based output, and the RF-based output is selected from: an indicator representing the alarm is faulty and a respective trigger-driven sensor, laser-based sensor, camera, and RF-based sensor is malfunctioning, an indicator representing no alarm and the respective trigger-driven sensor, laser-based sensor, camera, or RF-based sensor is properly functioning, or an indicator representing the alarm is valid and the respective trigger-driven sensor, laser-based sensor, camera, or RF-based sensor is properly functioning.
16. The system of claim 12, wherein the rule-based table defines a plurality of alarm classifications based on a combination of the trigger-driven output, the laser-based output, the image-based output, and the RF-based output.
17. The system of claim 12, wherein the rule-based table defines a plurality of alarm classifications based on a combination of the trigger-driven output, the laser-based output, and the image-based output.
18. The system of claim 12, wherein the rule-based table defines a plurality of alarm classifications based on a combination of the trigger-driven output, the laser-based output, and the RF-based output.
19. The system of claim 12, wherein the trigger-driven fault detection model is trained by: collecting historical trigger-driven sensor data from the environment; allocating a label to the historical trigger-driven sensor data based on an output of a historical alarm, the label is selected from: an indicator representing the historical alarm is faulty and a respective trigger-driven sensor is malfunctioning, an indicator representing no alarm and the respective trigger-driven sensor is properly functioning, or an indicator representing the historical alarm is valid and the respective trigger-driven sensor is properly functioning; and training the trigger-driven fault detection model by using the historical trigger-driven sensor data as an input and the label as a target output, wherein the trigger-driven fault detection model learns to correlate the input to the target output.
20. One or more non-transitory computer-readable media containing, in any combination, computer program code that, when executed by operation of a computer system, performs operations comprising: receiving an alarm triggered by a component within an environment monitored by a security system; generating one or more event outputs for the environment comprising at least three of: generating a trigger-driven output by processing trigger-driven data using an trigger-driven fault detection model, generating a laser-based output by processing laser-based data using a laser fault detection model, generating an image-based output by processing image-based data using an image fault detection model, or generating a radio frequency (RF)-based output by processing RF-based data using a RF fault detection model; comparing the one or more event outputs, comprising at least three of the trigger-driven output, the laser-based output, the image-based output, and the RF-based output, with a rule-based table; and generating a likelihood of the alarm being a false alarm based on the comparison.