Many of us have Smart Homes but are we actually smart about our homes? That is to say, while you may be able to interact with a variety of connected devices around your home, do you know when anything is on or off from anywhere? Or do you know how much energy a particular device you have is consuming? Or do you know how much energy is being consumed at your home all of the time? These are the very questions a new connected power device called Sense is trying to figure out, day and night. And Sense is using machine learning to do so. I recently install the Sense in my electrical panel and the insights I’m already getting are making me check the Sense app many times a day.
What’s really interesting is I got a preview of this type of technology back in 2010 as part of an Intel Influencers program. Deep in the Intel Labs, a few others and I saw how an electrical device could be “identified” by the electrical fingerprint it put out. Essentially, when any electrical device is powered on and running, it puts out a digital signature. You can think of it like a voice print that can not only identify what is being said, even contextually, but also who is saying it. The voice of a child saying “I want to go to the zoo” looks differently than an adult saying “the garage is dirty.” And even more precisely, one person saying “I went to the WHITE house” is contextually different than someone saying “I went to the white HOUSE.” These types of nuances can be differentiated in voice, pronunciation, emphasis, and other nuances.
Machine learning actually unlocks a lot of potential when it comes to interpreting huge sets of data. And this is exactly what Sense is trying to do. As more Sense devices are installed and start sucking in vast amounts of data, and that data is compared, it begins to identify patterns and signatures of electrical devices. The more data that is uploaded and analyzed, the smarter and more precise the data analytics becomes, and the more accurate the device identification also becomes.