Prevent false data from skewing machine learning
A new control option <machine learning> tab would be added to the device settings.
FOCI emotion detection is based on the minute ebb and flow motion of diaphragmatic breathing. Not only does it have to learn to differentiate real breathing from artifacts, it distinguishes the particularity of an individual’s breathing, and physiological features, to correctly read emotion states. And this learning in some situations COULD be skewed by false breathing data, such as:
Oscillation while dangling (e.g. on the usb charging cable)
Jiggling on shaky hard surface (e.g. on a table with unbalanced legs, on the table while you type on the keyboard)
Wobbling on unsteady soft surfaces (e.g. on the bed or sofa, where motion could be misread as real breathing)
And if the machine learning is skewed by these false data, there would be false streaks, emotion inference would consequently be skewed and you would have to either rebind the device to reset machine learning, in which case FOCI will relearn at faster pace, or wait for machine learning to slowly rectify itself, which would take longer.
Placing the device clip down would also stop machine learning from reading false breathing data
Now, with this option, you would have the choice of pausing machine learning when the device is in an environment that tends to generate false breathing signals. If you don’t usually see false streaks, then it wouldn’t affect you, and you need not worry about it at all.
The next few software updates, before the end of year, would revolve around closer integration of focus skill training. We are already testing mental performance approximation score and vicious loop detection within our team. It is pretty awesome and helpful. You will love it. We are teaching the machine to read emotion data, and it can do much better than a human already.
Mick & The FOCI Team
P.S. If you face any problems with shipment, message me at firstname.lastname@example.org and we’ll get it sorted.