This is the story of how I trained a simple neural network to solve a well-defined yet novel challenge in a real iOS app. The problem is unique, but most of what I cover should apply to any task in any iOS app.

That’s the beauty of neural networks. I’ll walk you through every step, from problem all the way to App Store.

On the way we’ll take a quick detour into an alternative approach using simple math (fail), through tool building, dataset generation, neural network architecting, and PyTorch training. We’ll endure the treacherous CoreML model converting to finally reach the React Native UI.

If this feels like too long a journey, not to worry. You can click the left side of this page to skip around.

And if you’re just looking for a tl;dr, here are some links:
code, test UI, iOS app, and my Twitter. I recently built a little iOS app for mechanical watch fanciers to track the accuracy of their watches over time. Read more from attardi.org…

thumbnail courtesy of attardi.org