Venkatasubramanian, Krishna Kumar
We demonstrate that by monitoring readings from the accelerometer and gyroscope of a head- mounted display, we are able to construct a waveform that is closely tied with the cardiac cycle of the wearer. Furthermore, we show that, from this waveform, we can then extract features that are not only consistent over time in the wearer, but also reasonably unique between different individuals. By then constructing an ensemble of random forest classifiers, we show that such a model can be used to determine if a new set of features does or does not belong to wearer. In this way, such a system can be used in an authentication context with a high degree of accuracy.
Worcester Polytechnic Institute
Major Qualifying Project