Training a machine-learned module for radar-based gesture detection in an ambient compute environment

A system comprising:

• a radar system;
• and
• a processor configured to process complex radar data generated by the radar system according to a machine-learned module having been trained with a two-phase evaluation process that adjusted one or more elements of the machine-learned module to reduce an error associated with classification of multiple gestures and to reduce a false positive rate, the two-phase evaluation process comprising:
• a segmented classification task that was performed using pre-segmented data to evaluate the error, the pre-segmented data comprising complex radar data having multiple gesture segments, each gesture segment of the multiple gesture segments comprising a gesture motion, each gesture segment of the multiple gesture segments representing a window of time that is centered based on a center of the gesture motion;
• and
• an unsegmented recognition task that was performed using continuous time-series data to evaluate the false positive rate, the continuous time-series data comprising other complex radar data corresponding to an absence of the multiple gestures.

A computer-readable storage medium comprising instructions that, responsive to execution by a processor, evaluates a machine-learned module using a two-phase evaluation process, the computer-readable storage medium configured to:

• perform, using pre-segmented data, a segmented classification task using the machine-learned module to evaluate an error associated with classification of multiple gestures, the pre-segmented data comprising complex radar data having multiple gesture segments, each gesture segment of the multiple gesture segments comprising a gesture motion, each gesture segment of the multiple gesture segments representing a window of time that is centered based on a center of the gesture motion;
• perform, using continuous time-series data, an unsegmented recognition task using the machine-learned module to evaluate a false positive rate, the continuous time-series data comprising other complex radar data corresponding to an absence of the multiple gestures;
• and
• adjust one or more elements of the machine-learned module to reduce the error and the false positive rate.