Abstract: The disclosed embodiments provide a system that identifies a user of an electronic device. During a training mode, the system uses an initial training data set, comprising sensor data from electronic devices associated with a set of initial users, to train a multilayer neural network model to authenticate the initial users. Next, the system uses an additional training data set, which includes sensor data from electronic devices associated with one or more new users, to update a portion of the weights in the trained model so that the updated model can be used to authenticate both the initial users and the one or more new users. During a subsequent surveillance mode, the system uses the updated model to authenticate a user of the electronic device based on sensor data contemporaneously received from the electronic device.

Abstract: The inventors recently developed a system that authenticates and/or identifies a user of an electronic device based on passive factors, which do not require conscious user actions. During operation of the system, in response to a trigger event, the system collects sensor data from one or more sensors in the electronic device, wherein the sensor data includes movement-related sensor data caused by movement of the portable electronic device while the portable electronic device is in control of the user. Next, the system extracts a feature vector from the sensor data, and analyzes the feature vector to authenticate and/or identify the user. During this process, the feature vector is analyzed using a model trained with sensor data previously obtained from the portable electronic device while the user was in control of the portable electronic device.

Abstract: The disclosed embodiments relate to a system that anonymizes sensor data to facilitate machine-learning training operations without disclosing an associated user's identity. During operation, the system receives encrypted sensor data at a gateway server, wherein the encrypted sensor data includes a client identifier corresponding to an associated user or client device. Next, the system moves the encrypted sensor data into a secure enclave. The secure enclave then: decrypts the encrypted sensor data; replaces the client identifier with an anonymized identifier to produce anonymized sensor data; and communicates the anonymized sensor data to a machine-learning system. Finally, the machine-learning system: uses the anonymized sensor data to train a model to perform a recognition operation, and uses the trained model to perform the recognition operation on subsequently received sensor data.

Abstract: The disclosed embodiments provide a system that authenticates a user of an unattended device. In response to sensing a presence of the user in proximity to the unattended device, the system makes a call from the unattended device to an authentication service to authenticate the user. In response to the call, the authentication service authenticates the user based on recently collected sensor data, which was obtained from one or more sensors in a portable electronic device belonging to the user. If authentication succeeds, the system allows the user to proceed with an interaction with the unattended device.

Abstract: The disclosed embodiments relate to a system that anonymizes sensor data to facilitate machine-learning training operations without disclosing an associated user's identity. During operation, the system receives encrypted sensor data at a gateway server, wherein the encrypted sensor data includes a client identifier corresponding to an associated user or client device. Next, the system moves the encrypted sensor data into a secure enclave. The secure enclave then: decrypts the encrypted sensor data; replaces the client identifier with an anonymized identifier to produce anonymized sensor data; and communicates the anonymized sensor data to a machine-learning system. Finally, the machine-learning system: uses the anonymized sensor data to train a model to perform a recognition operation, and uses the trained model to perform the recognition operation on subsequently received sensor data.

Abstract: The inventors recently developed a system that authenticates and/or identifies a user of an electronic device based on passive factors, which do not require conscious user actions. During operation of the system, in response to a trigger event, the system collects sensor data from one or more sensors in the electronic device, wherein the sensor data includes movement-related sensor data caused by movement of the portable electronic device while the portable electronic device is in control of the user. Next, the system extracts a feature vector from the sensor data, and analyzes the feature vector to authenticate and/or identify the user. During this process, the feature vector is analyzed using a model trained with sensor data previously obtained from the portable electronic device while the user was in control of the portable electronic device.

Abstract: The disclosed embodiments relate to a system that authenticates and/or identifies a user of an electronic device based on passive factors, which do not require conscious user actions. During operation of the system, in response to detecting a trigger event, the system collects sensor data from one or more sensors in the electronic device. Next, the system extracts a feature vector from the sensor data. The system then analyzes the feature vector to authenticate and/or identify the user, wherein the feature vector is analyzed using a model trained with sensor data previously obtained from the electronic device while the user was operating the electronic device.