Once you have created your account with Sense and your setup details are complete, Sense begins to collect data and run its device detection algorithms that rely on machine learning. Machine learning is both state-of-the-art and a technology that continues to improve as more data is collected.
Following is a simple overview of some frequently asked questions. We've included links that allow you to dig more deeply into the topics.
Frequently asked questions
What is Sense?
Sense is a technology based on machine learning algorithms. This means that Sense listens and gradually "learns" to differentiate unique patterns (think "signatures" or "fingerprints") of electrical devices used in your home. We then report this information so that your improved understanding can influence your choices in how to save energy and reduce costs.
How does Sense work?
Throughout the day, Sense analyzes changes in electrical magnitude, phase, and frequency. We then apply our algorithms to this data. Sense needs to “see” a device run a number of times before it can recognize, verify, and ultimately label a device's electrical signature. This all takes time. Not surprisingly, the first devices Sense identifies are likely to be common devices that cycle frequently and consume large amounts of power, such as your refrigerator or washing machine. Lower usage or smaller load devices like game consoles and televisions tend to take longer for Sense to find.
How long does device detection take?
You should begin to see devices detected within the first one to two weeks after Sense becomes active. Sense is always learning, and over the next several weeks (or months) it will continue to detect new devices and ask for you to name and verify its hypothesis.
How can I help with device detection?
As the Sense algorithm identifies devices and asks you to name them, you will want to participate in the process. It can take some trial and error. Read this blog for more insight into naming.
- Some devices are quite similar (think toaster and toaster oven), so Sense will benefit from your interaction.
- Some devices have more than one motor or heat source, so you may need to merge hypotheses if they actually refer to the same complex device (think of your dishwasher, which has 2-3 different mechanisms).
You may be tempted to “train” Sense, since you know your devices, but please let Sense lead the way. Here is a discussion of why trying to “train” won’t really work and could interfere with how Sense learns. This article discusses more of what you can do.
Will Sense find all my devices?
Due to variances across homes in terms of power quality and devices present, as well as frequenbcy of use, 100% detection is unlikely. However, Sense is always learning. Our detection algorithms will continue to improve as more homes work with Sense and more users do things like turning on networked device detection (article here) and renaming your detected mystery devices.
For more information on our state-of-the-art device detection systems, take a look at this blog article on the topic.