Presented By
O’Reilly + Cloudera
Make Data Work
March 25-28, 2019
San Francisco, CA

Talking to the machines: Monitoring production machine learning systems

Ting-Fang Yen (DataVisor)
5:10pm5:50pm Wednesday, March 27, 2019
Secondary topics:  Automation in data science and big data, Model lifecycle management, Temporal data and time-series analytics

Who is this presentation for?

  • Data scientists and engineers that build and maintain production machine learning models

Level

Intermediate

Prerequisite knowledge

  • A general understanding of machine learning technologies
  • Experience working with large-scale datasets and systems

What you'll learn

  • Understand the practical challenges of deploying machine learning systems
  • Learn how to maintain a production machine learning system that handles over a billion requests per day on average
  • Explore metrics for model quality when labeled data isn't available

Description

Production machine learning systems require constant monitoring, not just to keep the system online but also to ensure the model inference results are correct. This is much more straightforward when user feedback or labels are available. In those cases, the model performance can be tracked and periodically reevaluated using standard metrics such as precision, recall, or AUC. But what about when labeled data is lacking? In many applications, labels are expensive to obtain (requiring human analysts’ manual review) or cannot be obtained in a timely manner (e.g., not available until weeks or months later).

Ting-Fang Yen discusses the design and implementation of a real-time system to monitor production machine learning systems. The approach is designed to discover detection anomalies, such as volume spikes caused by spurious false positives, as well as gradual concept drifts when the model is no longer able to capture the target concept. In either case, it is able to automatically detect undesirable model behaviors early.

Part of the approach borrows from signal processing techniques for time series decomposition, where the time series can be used to represent a sequence of model decisions on different types of input data, or the amount of deviation between consecutive model runs. The approach calculates cross-correlation among the identified anomalies to facilitate root cause analysis of the model behavior.

This work is a step toward automated deployment of machine learning in production as well as new tools for interpreting model inference results.

Photo of Ting-Fang Yen

Ting-Fang Yen

DataVisor

Ting-Fang Yen is director of research at DataVisor, the leading fraud, crime, and abuse detection solution utilizing unsupervised machine learning to detect fraudulent and malicious activity such as fake account registrations, fraudulent transactions, spam, account takeovers, and more. She has over 10 years of experience in applying big data analytics and machine learning to tackle problems in cybersecurity. Ting-Fang holds a PhD in electrical and computer engineering from Carnegie Mellon University.

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)