Serverless Machine Learning: a beginner's guide
summary
A complete serverless pipeline that ingests images from Cloud Storage, converts pixels to columnar rows into BigQuery, batches exports to CSV, runs TensorFlow training on Cloud Machine Learning Engine and deploys an auto‑scaling prediction endpoint via Cloud Functions. A linear model hits ~92% accuracy, a deep network ~96%. The demo initially mislabels sevens because training lacked examples, then pulls live sevens, retrains and updates the deployed model successfully.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 1:15What is machine learning
- 3:45Why use machine learning
- 5:24Demo
- 8:05Mistake
- 8:28Training new models
- 9:29Serverless in production
- 11:21Machine learning workflow
- 12:23Tensorflow
- 14:24What is TensorFlow
- 19:49TensorFlow on other hardware
- 21:57Cloud Machine Learning Engine
- 22:36Endtoend
- 23:11Pieces of our application
- 26:49Google Cloud Products
- 28:50Cloud Functions
- 29:15Training
- 30:29Processing Data
- 32:13Serverless Steps
- 36:33Prediction
- 37:55Testing
- 39:09Resources
related talks
46mServerless Machine Learning: An Introduction
41mBuilding Serverless Machine Learning Models in the Cloud
40mServerless Machine Learning 101
38mServerless tools for machine learning, big data and app backends
12mServerless Deep Learning
26mDeploying Machine Learning Models in Serverless Architectures
43mMachine Learning at Scale
01h 27mDeploying a Machine Learning Model to the Cloud Using AWS Lambda