Access control to secured locations using relaxed biometrics
Example implementations include a method, apparatus, and computer-readable medium for controlling access to a location, comprising receiving an access identifier and a biometric information sample from a user attempting to enter an access point. The implementations further include matching the access identifier to an authenticated cluster identifier corresponding to a biometric information cluster within a plurality of different biometric information clusters, wherein the biometric information cluster is associated with an authenticated user. Additionally, the implementations further include identifying, based on a machine learning model processing, a cluster identifier from the biometric information sample. Additionally, the implementations further include transmitting a first signal to allow access to the location in response to the inferred biometric information cluster identifier matching an authenticated cluster identifier, or a second signal to deny access to the location in response to the inferred biometric information cluster identifier not matching an authenticated cluster identifier.
1. A method for controlling access to a location, comprising:
receiving an access identifier and a biometric information sample from a user attempting to enter an access point;
matching the access identifier to an authenticated cluster identifier corresponding to a biometric information cluster within a plurality of different biometric information clusters, wherein the biometric information cluster is associated with an authenticated user, and wherein the biometric information cluster includes only non-unique biometric information associated with a plurality of users;
identifying, based on a machine learning model processing the biometric information sample, a cluster identifier of an inferred biometric information cluster within the plurality of different biometric information clusters associated with the biometric information sample; and
transmitting a first signal to allow access to the location in response to the cluster identifier of the inferred biometric information cluster matching the authenticated cluster identifier, or a second signal to deny access to the location in response to the cluster identifier of the inferred biometric information cluster not matching the authenticated cluster identifier.
2. The method of claim 1 , wherein the machine learning model is a person identification model, and wherein the non-unique biometric information associated with the plurality of users comprises a first set of information at an intermediate layer of the person identification model prior to a second set of information that uniquely identifies the user in a final layer of the person identification model.
3. The method of claim 1 , wherein a number of the plurality of different biometric information clusters is configured based on a personal identification information restriction level.
4. The method of claim 1 , wherein a number of the plurality of different biometric information clusters is configured based on a confidence threshold.
5. The method of claim 1 , wherein a number of the plurality of different biometric information clusters is configured based on a personal identification information restriction level and a confidence threshold.
6. The method of claim 1 , wherein the biometric information sample comprises a voice sample, and wherein the machine learning model comprises a speaker recognition model.
7. The method of claim 1 , wherein the biometric information sample comprises a plurality of different types of biometric information from the user, and wherein identifying the cluster identifier of the inferred biometric information cluster within the plurality of different biometric information clusters comprises processing each of the plurality of different types of biometric information with a type-specific one of a plurality of different types of machine learning models.
8. The method of claim 7 , wherein at least two of the plurality of different biometric information clusters have different confidence thresholds.
9. The method of claim 1 , further comprising:
requesting an additional biometric information sample in response to the cluster identifier of the inferred biometric information cluster not matching the authenticated cluster identifier;
performing a second identifying, based on the machine learning model processing the additional biometric information sample, of a second cluster identifier of a second inferred biometric information cluster within the plurality of different biometric information clusters associated with the additional biometric information sample;
wherein transmitting the first signal to allow access to the location is in response to the second cluster identifier of the second inferred biometric information cluster matching the authenticated cluster identifier; and
wherein transmitting the second signal to deny access to the location is in response to the second cluster identifier of the second inferred biometric information cluster not matching the authenticated cluster identifier.
10. The method of claim 9 , wherein the additional biometric information sample is a different type of biometric information than the biometric information sample.
11. The method of claim 1 , further comprising:
collecting an initial biometric information sample from the authenticated user;
processing the initial biometric information sample by the machine learning model to identify the authenticated cluster identifier of the biometric information cluster within the plurality of different biometric information clusters associated with the authenticated user; and
storing the authenticated cluster identifier in a relationship with the access identifier associated with the authenticated user.
12. An apparatus for controlling access to a location, comprising:
one or more memories; and
one or more processors communicatively coupled with the one or more memories and, individually or in combination, configured to:
receive an access identifier and a biometric information sample from a user attempting to enter an access point;
match the access identifier to an authenticated cluster identifier corresponding to a biometric information cluster within a plurality of different biometric information clusters, wherein the biometric information cluster is associated with an authenticated user, and wherein the biometric information cluster includes only non-unique biometric information associated with a plurality of users;
identify, based on machine learning model processing the biometric information sample, a cluster identifier of an inferred biometric information cluster within the plurality of different biometric information clusters associated with the biometric information sample; and
transmit a first signal to allow access to the location in response to the cluster identifier of the inferred biometric information cluster matching the authenticated cluster identifier, or a second signal to deny access to the location in response to the cluster identifier of the inferred biometric information cluster not matching the authenticated cluster identifier.
13. The apparatus of claim 12 , wherein the machine learning model is a person identification model, and wherein the non-unique biometric information associated with the plurality of users comprises a first set of information at an intermediate layer of the person identification model prior to a second set of information that uniquely identifies the user in a final layer of the person identification model.
14. The apparatus of claim 12 , wherein a number of the plurality of different biometric information clusters is configured based on a personal identification information restriction level.
15. The apparatus of claim 12 , wherein a number of the plurality of different biometric information clusters is configured based on a confidence threshold.
16. The apparatus of claim 12 , wherein a number of the plurality of different biometric information clusters is configured based on a personal identification information restriction level and a confidence threshold.
17. The apparatus of claim 12 , wherein the biometric information sample comprises a voice sample, and wherein the machine learning model comprises a speaker recognition model.
18. The apparatus of claim 12 , wherein the biometric information sample comprises a plurality of different types of biometric information from the user, and wherein to identify the cluster identifier of the inferred biometric information cluster within the plurality of different biometric information clusters comprises processing each of the plurality of different types of biometric information with a type-specific one of a plurality of different types of machine learning models.
19. The apparatus of claim 18 , wherein at least two of the plurality of different biometric information clusters have different confidence thresholds.
20. The apparatus of claim 12 , wherein the one or more processors are further configured to:
request an additional biometric information sample in response to the cluster identifier of the inferred biometric information cluster not matching the authenticated cluster identifier;
perform a second identifying, based on the machine learning model processing the additional biometric information sample, of a second cluster identifier of a second inferred biometric information cluster within the plurality of different biometric information clusters associated with the additional biometric information sample;
wherein to transmit the first signal to allow access to the location is in response to the second cluster identifier of the second inferred biometric information cluster matching the authenticated cluster identifier; and
wherein to transmit the second signal to deny access to the location is in response to the second cluster identifier of the second inferred biometric information cluster not matching the authenticated cluster identifier.
21. The apparatus of claim 20 , wherein the additional biometric information sample is a different type of biometric information than the biometric information sample.
22. The apparatus of claim 12 , wherein the one or more processors are further configured to:
collect an initial biometric information sample from the authenticated user;
process the initial biometric information sample by the machine learning model to identify the authenticated cluster identifier of the biometric information cluster within the plurality of different biometric information clusters associated with the authenticated user; and
store the authenticated cluster identifier in a relationship with the access identifier associated with the authenticated user.