User identification via motion and heartbeat waveform data
First Claim
Patent Images
1. A method, implemented on a wearable fitness monitor comprising one or more motion sensors and one or more processors, the method comprising:
- (a) obtaining motion data of a wearer from the one or more motion sensors when the wearable fitness monitor is worn by the wearer, wherein the wearable fitness monitor is configured to be worn by a person;
(b) obtaining, by the one or more processors, a motion signature using the motion data of the wearer, wherein the motion signature characterizes a body movement;
(c) obtaining, by the one or more processors, a test feature vector from the motion signature;
(d) applying, by the one or more processors, a machine learning classifier to the test feature vector, whereinthe machine learning classifier was trained using motion data reflecting motions of one or more non-humans or one or more humans,the machine learning classifier is configured to receive feature vectors as inputs and provide classifications of the feature vectors as outputs, andthe classifications indicate whether the feature vectors are obtained from non-humans;
(e) obtaining, using the machine learning classifier and by the one or more processors, a classification indicating whether the test feature vector is obtained from a non-human;
(f) determining, based on the classification obtained in (e) and by the one or more processors, the motion signature corresponds to an invalid motion feature, the invalid motion feature characterizing motion likely to be performed by a non-human; and
(g) preventing, based on the determination in (f) and by the one or more processors, the wearable fitness monitor from allowing a transaction.
1 Assignment
0 Petitions
Accused Products
Abstract
The disclosure relates to methods, devices, and systems to identify a user of a wearable fitness monitor using data obtained using the wearable fitness monitor. Data obtained from motion sensors of the wearable fitness monitor and data obtained from heartbeat waveform sensors of the wearable fitness monitor may be used to identify the user.
-
Citations
20 Claims
-
1. A method, implemented on a wearable fitness monitor comprising one or more motion sensors and one or more processors, the method comprising:
-
(a) obtaining motion data of a wearer from the one or more motion sensors when the wearable fitness monitor is worn by the wearer, wherein the wearable fitness monitor is configured to be worn by a person; (b) obtaining, by the one or more processors, a motion signature using the motion data of the wearer, wherein the motion signature characterizes a body movement; (c) obtaining, by the one or more processors, a test feature vector from the motion signature; (d) applying, by the one or more processors, a machine learning classifier to the test feature vector, wherein the machine learning classifier was trained using motion data reflecting motions of one or more non-humans or one or more humans, the machine learning classifier is configured to receive feature vectors as inputs and provide classifications of the feature vectors as outputs, and the classifications indicate whether the feature vectors are obtained from non-humans; (e) obtaining, using the machine learning classifier and by the one or more processors, a classification indicating whether the test feature vector is obtained from a non-human; (f) determining, based on the classification obtained in (e) and by the one or more processors, the motion signature corresponds to an invalid motion feature, the invalid motion feature characterizing motion likely to be performed by a non-human; and (g) preventing, based on the determination in (f) and by the one or more processors, the wearable fitness monitor from allowing a transaction. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13)
-
-
14. A system comprising:
-
a wearable fitness monitor configured to be worn by a person and comprising; one or more motion sensors, and a communication interface configured for communicating data from the one or more motion sensors to a device external to the wearable fitness monitor; and logic configured to; (a) obtain motion data of a wearer from the one or more motion sensors when the wearable fitness monitor is worn by the wearer, wherein the wearable fitness monitor is configured to be worn by a person; (b) obtain a motion signature using the motion data of the wearer, wherein the motion signature characterizes a body movement; (c) obtain a test feature vector from the motion signature; (d) apply a machine learning classifier to the test feature vector, wherein the machine learning classifier was trained using motion data reflecting motions of one or more non-humans or one or more humans, the machine learning classifier is configured to receive feature vectors as inputs and provide classifications of the feature vectors as outputs, and the classifications indicate whether the feature vectors are obtained from non-humans; (e) obtain, using the machine learning classifier, a classification indicating whether the test feature vector is obtained from a non-human; (f) determine, based on the classification obtained in (e), the motion signature corresponds to an invalid motion feature, the invalid motion feature characterizing motion likely to be performed by a non-human; and (g) prevent, based on the determination in (f), the wearable fitness monitor from allowing a transaction. - View Dependent Claims (15, 16, 17, 18, 19, 20)
-
Specification