If you hear the phrase ‘on-device machine-learning’, it refers to technology that operates entirely on a device – for example a smartphone – rather than passing data back to a company’s servers for processing. One of the main benefits: privacy, because the device owner’s personal data is staying on their personal device.
This is all part of the pitch for a new startup called Canopy, which “uses on-device machine learning to understand what you like in order to predict what you’ll love” – including music, but also news articles, places you’ve visited. Canopy will send people “a handful of new recommendations” each day to “excite, inspire, and motivate you” – but it WON’T be building up a profile on its servers of anyone who’s using the service. The interesting part? Former Spotify execs are behind the company: including The Echo Nest’s co-founder and CTO Brian Whitman, and former Spotify director of music publishing Annika Goldman. See Whitman’s quote (in a TechCrunch interview) about his time at Spotify, and “how much power we had put in algorithmic decision making and personalisation. All your information goes to their servers”. Canopy is backed by $4.5m of seed funding.