Smoothing of discretized values using a transition matrix
Techniques are provided for smoothing discretized values used, for example, for authentication or identity assurance. An illustrative method comprises obtaining at least one probability of transitioning between at least two discretized values of a given feature; computing a smoothed feature score for the given feature for a transition from a first one of the discretized values to a second one of the discretized values based on the probability of the transition from the first discrete value to the second discrete value; and performing one or more of authenticating a user and verifying an identity of a user based at least in part on the smoothed feature score. The probabilities of transitioning between the discretized values are optionally stored in a transition matrix. Feature scores for first and second discretized values are optionally weighted based on the probability of the transition from the first discrete value to the second discrete value.
1. A method, comprising:
obtaining at least one probability of transitioning between at least two discretized values of a given feature, wherein the at least two discretized values are a result of a discretization procedure such that each of the at least two discretized values represents a range of two or more values corresponding to the given feature, wherein each of the at least two discretized values are associated with a respective feature score;
computing, using at least one processing device, a smoothed feature score for said given feature for a transition from a first one of said discretized values to a second one of said discretized values based at least in part on: (i) the feature score associated with the first discretized value, (ii) the feature score associated with the second discretized value, and (iii) the probability of the transition from the first discretized value to the second discretized value, wherein computing the smoothed feature score smoothes common transitions and does not smooth uncommon transitions; and
performing, using said at least one processing device, one or more of authenticating a user and verifying an identity of a user based at least in part on said smoothed feature score.
2. The method of claim 1 , further comprising the step of detecting when said given feature transitions from said first one of said discretized values to said second one of said discretized values.
3. The method of claim 1 , wherein said probabilities of transitioning between said at least two discretized values are stored in a transition matrix.
4. The method of claim 1 , wherein said step of computing a smoothed feature score for said given feature is not performed if the probability of the transition from the first discretized value to the second discretized value is below a predefined threshold value.
5. The method of claim 1 , further comprising the step of aggregating the smoothed feature score for the given feature with one or more feature scores for one or more additional features to obtain an aggregate score, and wherein said performing is further based on said aggregated score.
6. The method of claim 1 , wherein said smoothed feature score for said given feature for said transition from said first one of said discretized values to said second one of said discretized values is computed by weighting the feature score associated with said first one of said discretized values based on said probability of the transition from the first discretized value to the second discretized value and weighting the feature score associated for said second one of said discretized values based on said probability of the transition from the first discretized value to the second discretized value.
7. The method of claim 6 , wherein said weighting employs one or more of a weighted average and an exponential decay.
8. The method of claim 1 , wherein computing the smoothed feature score is based on a direction of user data corresponding to said given feature.
9. A computer program product, comprising a tangible machine-readable storage medium having encoded therein executable code of one or more software programs, wherein the one or more software programs when executed by at least one processing device perform the following steps:
obtaining at least one probability of transitioning between at least two discretized values of a given feature, wherein the at least two discretized values are a result of a discretization procedure such that each of the at least two discretized values represents a range of two or more values corresponding to the given feature, wherein each of the at least two discretized values are associated with a respective feature score;
computing, using at least one processing device, a smoothed feature score for said given feature for a transition from a first one of said discretized values to a second one of said discretized values based at least in part on: (i) the feature score associated with the first discretized value, (ii) the feature score associated with the second discretized value, and (iii) the probability of the transition from the first one of the discretized values to the second one of the discretized values, wherein computing the smoothed feature score smoothes common transitions and does not smooth uncommon transitions; and
performing, using said at least one processing device, one or more of authenticating a user and verifying an identity of a user based at least in part on said smoothed feature score.
10. The computer program product of claim 9 , wherein said probabilities of transitioning between said at least two discretized values are stored in a transition matrix.
11. The computer program product of claim 9 , wherein said step of computing a smoothed feature score for said given feature is not performed if the probability of the transition from the first discretized value to the second discretized value is below a predefined threshold value.
12. The computer program product of claim 9 , further comprising the step of aggregating the smoothed feature score for the given feature with one or more feature scores for one or more additional features to obtain an aggregate score, and wherein said performing is further based on said aggregated score.
13. The computer program product of claim 9 , wherein said smoothed feature score for said given feature for said transition from said first one of said discretized values to said second one of said discretized values is computed by weighting the feature score associated with said first one of said discretized values based on said probability of the transition from the first discretized value to the second discretized value and weighting the feature score associated with said second one of said discretized values based on said probability of the transition from the first discretized value to the second discretized value.
14. The computer program product of claim 13 , wherein said weighting employs one or more of a weighted average and an exponential decay.
15. An apparatus, comprising:
a memory; and
at least one processing device, coupled to the memory, operative to implement the following steps:
obtaining at least one probability of transitioning between at least two discretized values of a given feature, wherein the at least two discretized values are a result of a discretization procedure such that each of the at least two discretized values represents a range of two or more values corresponding to the given feature, wherein each of the at least two discretized values are associated with a respective feature score;
computing, using said at least one processing device, a smoothed feature score for said given feature for a transition from a first one of said discretized values to a second one of said discretized values based at least in part on: (i) the feature score associated with the first discretized value, (ii) the feature score associated with the second discretized value, and (iii) the probability of the transition from the first discretized value to the second discretized value; and
performing, using said at least one processing device, one or more of authenticating a user and verifying an identity of a user based at least in part on said smoothed feature score.
16. The apparatus of claim 15 , wherein said probabilities of transitioning between said at least two discretized values are stored in a transition matrix.
17. The apparatus of claim 15 , wherein said step of computing a smoothed feature score for said given feature is not performed if the probability of the transition from the first discretized value to the second discretized value is below a predefined threshold value.
18. The apparatus of claim 15 , further comprising the step of aggregating the smoothed feature score for the given feature with one or more feature scores for one or more additional features to obtain an aggregate score.
19. The apparatus of claim 15 , wherein said smoothed feature score for said given feature for said transition from said first one of said discretized values to said second one of said discretized values is computed by weighting a feature score for said first one of said discretized values based on said probability of the transition from the first discretized value to the second discretized value and weighting a feature score for said second one of said discretized values based on said probability of the transition from the discretized discrete value to the second discretized value.
20. The apparatus of claim 19 , wherein said weighting employs one or more of a weighted average and an exponential decay.