Explore testing and deployment strategies for machine learning projects in this comprehensive lecture from the Full Stack Deep Learning March 2019 bootcamp. Delve into topics such as project structure, AB testing, evaluation methods, continuous integration, and software services. Learn about deployment options including virtual machines, Docker containers, REST APIs, and serverless architectures. Discover best practices for load balancing, dependency management, and handling distribution shifts. Gain insights on CPU-only, batch, and algorithmic deployment techniques, as well as strategies for rollbacks and optimizing startup times.
Testing and Deployment - Full Stack Deep Learning - March 2019
Overview
Syllabus
Introduction
Outline
Project Structure
Machine Learning
AB Testing
Evaluation Tests
Research
Software Engineering
Validation
Distribution shifts
Continuous integration and testing
Software services
Virtual Machine
Docker Container
Dockerfile
Docker Hub
REST API
Prediction System
Deployment Options
Load Balancer
Dependency
Serverless
rollback
startup time
CPU only deployment
Batch deployment
Algorithmic deployment
Taught by
The Full Stack
