Training a Neural Network to Fall

The article’s title, in the formal sense, is a bit misleading, as the neural network's purpose is for the detection (and not simulation) of falls.

IMO an interesting part in the article which is in regards to the training data is: "the client’s data science firm used stunts people to fall like elderly people while wearing the sensors and streaming the collected data..".

So this is pretty speculative, but I think Google already has similar models in production that they use for targeted advertisements.

A few months ago I was riding my bike and fell down. A few hours later while looking at my Google Pixel I saw an advertisement for a nearby medical clinic. The ad was a guy on the ground holding his knee after a fall. It could be a coincidence, but I have never seen any similar ads before or after. It occurred to me that it wouldn’t be too difficult to build a fall classifier based on the phone’s accelerometer and gyroscope data.

Who would have thought that choosing an LSTM on iteration 1 for embedded C code would have been fraught with challenges.