This repository contains a SWIFTUI implementation of a LSTM. This is the baseline Machine Learning algorithm used in the Skitag App. It differs in terms of optimization from the one that is installed in the latest version of Skitag. We publish this algorithm with the intention to share this library to anyone who has curiosity about both SWIFTUI and LSTM, and we do it also to be compliant with Skitag's AI Ethics commitment to transparency.
You can find additional information about Skitag in: https://skitag.eu
Unless you are a big fun of coding and you are really a curious mind, we highly recommend to use Core ML for your AI implementations. It is much more easy and straightforward; e.g. MLActivityClassifier.
Nevertheless, in case you (like we) need to train a ML from scratch, and do it on-device, this is the right repository for your. At Skitag we choose this path because our dream is to let you have your own customized AI trained in your own device, and from scratch. Enjoy it!