Abstract
A machine learning system is described that enables a device (e.g., a wearable device or a mobile device) to personalize gestures for a user of the device based on movements of the user. Various motion sensors of the device may generate motion data that represents movements of the device. The device may utilize machine learning models to process the motion data to determine whether the user has performed a particular gesture (e.g., a tilt-to-wake gesture, hereinafter “TTW”). If the device determines that the user did preform the particular gesture, the device may perform an action corresponding to the particular gesture (e.g., the device may activate a display of the device in response to the user performing a tilt-to-wake gesture). The device may determine whether the determination of gesture performance was accurate (e.g., did the use interact with the device after the display was activated, which would indicate that the user intended to perform the tilt-to-wake gesture). If the device determines that the determination of gesture performance was accurate, the device may utilize the motion data to further train the machine learning models to process subsequent motion data. In this way, the device may personalize machine learning models to a user.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Freeman, Tyler, "TRAINING FOR MACHINE LEARNING MODELS TO PERSONALIZE GESTURES IN WEARABLE AND MOBILE DEVICES", Technical Disclosure Commons, (April 05, 2019)
https://www.tdcommons.org/dpubs_series/2122