Communications of the ACM

This installment of Research for Practice features a curated selection from Dan Crankshaw and Joey Gonzalez, who provide an overview of machine learning serving systems. What happens when we wish to actually deploy a machine learning model to production, and how do we serve predictions with high accuracy and high computational

efficiency? Dan and Joey’s picks provide a thoughtful selection of cutting-edge techniques spanning database-level integration, video processing, and prediction middleware. Given the explosion of interest in machine learning and its increasing impact on seemingly every application vertical, it is possible that systems such as these will become as commonplace as relational databases are today. Enjoy your read!

—Peter Bailis

Peter Bailis is an assistant professor of computer science at Stanford University. His research in the Future Data Systems group ( futuredata.stanford.edu) focuses on the design and implementation of next-generation data-intensive systems.

Machine learning is an enabling technology that transforms data into solutions by extracting patterns that generalize to new data. Much of machine learning can be reduced to learning a model—a function that maps an input (for example, a photo) to a prediction (for example, objects in the photo). Once trained, these models can be used to make predictions on new inputs (for example, new photos) and as part of more complex decisions (for example, whether to promote a photo). While thousands of papers are published each year on how to design and train models, there is surprisingly less research on how to