Getting Started with Machine Learning in your iOS apps

on YouTube ↗25m

" summary

Overview of how to ship on-device ML: use a pre-trained model (commonly TensorFlow Inception V3) exported and translated into Swift/Obj‑C, load it into Apple’s frameworks, then run inference with MPS/Metal GPU inference or BNNS (CPU). Example shows converting ~30 binary dat files into a Swift model and a sample app that classifies a bike with 83.35% mountain bike prediction. Measure impact against a KPI (e.g. listening time hypothesized from 30 to 38 minutes) and A/B test results.

AI-written summary · about

" chapters (6)

1. 0:00Intro
2. 0:15Getting Started with Machine Learning in your iOS Apps App Builders
3. 5:54Crash Course * ML Fundamentals
4. 11:34Where do I get the model?
5. 13:37Metal Performance Shader Convolutional Neural Networks
6. 20:04Why do inference on E? → Unique selling points (USPs) of ML on iOS device 1. Data privacy 2. Free computing power o

" related talks

• 40mDeep Learning on iOSShuichi Tsutsumi · 360iDev 2017
• 40mNavigating the Machine Learning Landscape for Apple DevicesMeghan Kane · Swift & Fika 2018
• 27mMachine Learning and MobileBrett Koonce · Scale By The Bay 2019
• 22mCore ML and Your AppMugunth Kumar · try! Swift India 2017
• 31mPower-Up Your App With Apple’s Machine Learning FrameworksKim Foong Chow · iOS Conf SG 2017
• 41mTensorFlow Neural Networks on iOSTaylan Pince · 360iDev 2017
• 22mGetting Started with Apple MLAudrey Tam · try! Swift NYC 2018
• 34mDemystifying ML in iOSSanfilippo, Malesani · Swift Heroes 2018