Building Machine Learning Pipelines
A machine learning project usually includes many moving parts, such as data processing, model training, model inference and model deployment. Because of the iterative and exploratory nature of developing a machine learning project, using a pipeline can make development faster and more effective.
Similar to the ETL process for data, machine learning projects can also have a development pipeline that pre-defines high-level building blocks. These building blocks work to establish a clear structure for the machine learning workflows. Through using pipelines, a large amount of the machine learning workflow can be automated, improvements to the models performance can be tracked and collaboration between engineers is simplified.
This talk will discuss some high-level building blocks in pipelines for ML projects: the concepts of DAG (Directed Acyclic Graph) and how it relates to machine learning pipelines; experience of building ML pipelines for our current project, with light-weight framework like `consecution` in Python and comprehensive framework like `Tensorflow Extended` with `Airflow`.
Target Audience
Machine Learning Engineers / People who have interest in machine learning.
schedule Submitted 1 year ago
People who liked this proposal, also liked:
-
keyboard_arrow_down
Effy Elden - False Assumptions Developers Make About Gender (and their sometimes hilarious results)
30 Mins
Talk
Beginner
From signup forms to social media profiles to targeted advertising: gender is everywhere.
But too often software developers make assumptions about gender that just don't work in the real world, and when those assertions fail, the results can be anywhere from amusing to seriously harmful.
Watch and learn as a nonbinary person breaks the gender binary, challenges your assumptions, and demonstrates the hilarious consequences of being an X in an M/F world.
-
keyboard_arrow_down
Brett Morgan - Flutter Everywhere
30 Mins
Talk
Intermediate
Flutter is a UI Toolkit by Google focused on empowering developers to build building beautiful, natively compiled applications quickly. Up until now, Flutter has focused on iOS and Android development, but is in the process of expanding its applicability, and empowering developers to make reactive user interfaces everywhere from Mobile, to the Web and Desktop. In this talk we will review, live coding style, all the places you can use Flutter.