AIDevFest20: Machine Learning Design Patterns for MLOps

preview_player
Показать описание
Speaker: Valliappa Lakshmanan (LAK), Google
Abstract: Design patterns are formalized best practices to solve common problems when designing a software system. As machine learning moves from being a research discipline to a software one, it is useful to catalog tried-and-proven methods to help engineers tackle frequently occurring problems that crop up during the ML process. In this talk, I will cover five patterns (Workflow Pipelines, Transform, Multimodal Input, Feature Store, Cascade) that are useful in the context of adding flexibility, resilience and reproducibility to ML in production. For data scientists and ML engineers, these patterns provide a way to apply hard-won knowledge from hundreds of ML experts to your own projects.
Anyone designing infrastructure for machine learning will have to be able to provide easy ways for the data engineers, data scientists, and ML engineers to implement these, and other, design patterns.
Рекомендации по теме
Комментарии
Автор

Starts at 02:40
Workflow pipeline 05:10
Transform 12:40
Multimodal input 18:10
Feature store 23:39
Cascade 28:30
Q&A 35:00

LakshmananValliappa
Автор

I don't see the point of Transform pattern - a standard 21st century API documentation for AI service will have all the features (inputs) specification. And as clients who are trying to opt-in for code/low-code solution - I will have no choice but to follow the spec before making inference calls. Perhaps I am missing something

haua
Автор

I don't agree that features transforms should be incorporated into a model itself, for instance, via tf Lambda layer. What if a DS wanted to switch from ANNs to boosting models? But if feature engineering step is managed by a separate pipeline, this won't be a problem...

anatolyalekseev
visit shbcf.ru