Even without native detections, Sense can still prove incredibly useful towards saving money in your home. The Power Meter is a fantastic tool that provides a real-time view of your energy consumption. If you want to know how much a device in your home is consuming, but Sense hasn’t discovered it yet, just open up the Power Meter and turn it off the next time it runs. You should see a big drop. You can do the same for the “Always On” devices in your home, identifying how much they’re costing you every day.
Still, our users frequently ask if there’s any data they can provide to Sense that will help detection improve. While you cannot directly “teach” or “train” Sense to find your devices, there are still quite a few ways you can assist in the discovery process.
Network Identification allows Sense to see some of the simple “handshake” messages put out by your networked devices (think smart TVs and refrigerators). While we can’t tell you if your milk is going bad or if your kids are watching too many action movies, this helps Sense to see when these devices are running and can thus help to correlate their run times with consumption patterns.
Some smart devices give Sense instant detections, like smart bulbs from Philips Hue and smart plugs from TP-Link Kasa and Belkin Wemo. This is great for you, but it also feeds Sense some fantastic data, helping to improve device detection for everybody.
As Sense continues to find your devices, make sure you’re providing make/model information and renaming them if needed (the Community Names feature is great for this). This also provides the team useful data to improve device models globally. In some cases, Sense may even take these steps automatically. Be sure to also fill out your Home Details, as it provides us a useful inventory of the types of devices in your home that we can use to build out better detection models.
When Sense finds a device, but you’re finding the detections to be inaccurate, you can report it as “not on.” This feeds our Data Science team valuable information so they can continue to refine the detection model. While you likely won’t see instant results from this, this reporting will help the entire Sense userbase into the future.
Please note that Sense needs to see devices in their usual context with their regular usage pattern to be able to accurately identify them. That means that “training” Sense by turning devices on/off and labeling them is not an effective method to help Sense learn. You can learn more about the challenges of implementing a “learning” mode here.
If you want to learn more about the technology behind Sense, check out our Machine Learning explainer video.