Analyzing and deploying your machine learning model
If you’ve invested time into collecting data, validating it, and training a machine learning model, you want that model to be used by other people. For this to happen, it’s essential to have a smooth pipeline from the input data to the deployed model in a production system. Catherine Nelson and Hannes Hapke focus on the key steps of this pipeline that follow training the model: analyzing the model’s predictions and deploying the model to production.
You’ll identify useful metrics for understanding your model in production, including those that may allow you to uncover biases in the model’s predictions. Catherine and Hannes explain how to do this using TensorFlow Model Analysis and the Google What-If Tool and offer an overview of tools for interpreting and explaining the predictions.
You’ll learn to deploy your machine learning model efficiently. After outlining the limitations of a simple Flask implementation, Catherine and Hannes introduce TensorFlow Serving and highlight its advantages, including its advanced batching and monitoring functionality. They guide you through simple and seamless model deployments.
Throughout, you’ll see a demonstration of the steps using an example project based on open source data. And you’ll be able to apply the content immediately to your data science problems. The content is available as a public GitHub repo for you to refer to afterwards.
- Familiarity with basic machine learning concepts and at least one machine learning framework (e.g., PyTorch, TensorFlow, Keras)
Materials or downloads needed in advance
- A laptop with Python 3, a Git client, and Docker client installed (You will have access to a public GitHub repo to download.)
What you'll learn
- Learn the necessary steps to deploy trained models to production, how to analyze models for prediction bias, how to use TensorFlow Model Analysis and the Google What-If Tool for model analysis, how to deploy models with TensorFlow Serving, how to configure TensorFlow Serving for model deployment, how to optimize models for deployment, and how to monitor the model deployment
Concur Labs, SAP Concur
Catherine Nelson is a senior data scientist for Concur Labs at SAP Concur, where she explores innovative ways to use machine learning to improve the experience of a business traveller. She’s particularly interested in privacy-preserving ML and applying deep learning to enterprise data. Previously, she was a geophysicist and studied ancient volcanoes and explored for oil in Greenland. Catherine has a PhD in geophysics from Durham University and a master’s of earth sciences from Oxford University.
Hannes Hapke is a senior data scientist at SAP ConcurLabs. He’s been a machine learning enthusiast for many years and is a Google Developer Expert for machine learning. Hannes has applied deep learning to a variety of computer vision and natural language problems, but his main interest is in machine learning infrastructure and automating model workflows. Hannes is a coauthor of the deep learning publication Natural Language Processing in Action and he’s currently working on a book about TensorFlow Extended Building Machine Learning Pipelines (O’Reilly). When he isn’t working on a deep learning project, you’ll find him outdoors running, hiking, or enjoying a good cup of coffee with a great book.
Leave a Comment or Question
Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?
Join the conversation here (requires login)
Premier Diamond Sponsors
Premier Exhibitor Plus
For conference registration information and customer service
For more information on community discounts and trade opportunities with O’Reilly conferences
For information on exhibiting or sponsoring a conference
For media/analyst press inquires