Palestra: Machine Learning Design Patterns

Track: Engenharia aplicada a Machine Learning

Sala: Sala 4

Horário: 10:50am - 11:35am

Dia da semana: Segunda-feira

Nível: Intermediário

Persona: Arquiteto(a), Cientista de Dados, Desenvolvedor(a) Sênior

Apresentação em Inglês

Share this on:

Pontos Principais

  • Moving an ML model to production is much easier if you keep inputs, features, and transforms separate
  • Saving the intermediate weights of your model during training provides resilience, generalization, and tunability
  • Base machine learning model training and evaluation on the total number of examples, not on epochs or steps
  • Export your model so that it passes through client keys

Resumo

As the practice of Machine Learning gets formalized, the community learns best practices of setting up large scale training loops and moving from development to production. In the talk, I introduce some of these design patterns, explaining the problem and the code for these patterns in Keras and Tensorflow 2.0.

Palestrante: Valliappa Lakshmanan

Head of Data Analytics and AI Solutions, Google Cloud

Lak is Global Head for Data Analytics and AI Solutions on Google Cloud. His team builds software solutions for business problems using Google Cloud's data analytics and machine learning products. He founded Google's Advanced Solutions Lab ML Immersion program and is the author of a couple of O'Reilly books and several Coursera courses. Before Google, Lak was a Director of Data Science at Climate Corporation and a Research Scientist at NOAA.

Find Valliappa Lakshmanan at

Tracks 2019