Presented By O'Reilly and Cloudera
Make Data Work
March 28–29, 2016: Training
March 29–31, 2016: Conference
San Jose, CA

How good is your model? New tools for reliable data science

Moritz Hardt (Google)
9:05am–9:30am Tuesday, 03/29/2016
Hardcore Data Science
Location: 210 C/G
Average rating: ****.
(4.31, 13 ratings)

Prerequisite knowledge

Attendees should be familiar with data science practices.


Most applications of machine learning across science and industry rely on the holdout method for model selection and validation. Unfortunately, the holdout method can fail in the now common situation
where the data scientist works interactively with the data, iteratively choosing which methods to use by probing the same holdout data many times.

Moritz Hardt outlines a reusable holdout method, which can be used many times without losing the guarantees of fresh data. Moritz also explains how to design reliable machine-learning benchmarks for a number of applications such as data science competitions and hyperparameter tuning.

Photo of Moritz Hardt

Moritz Hardt


Moritz Hardt is a senior research scientist at Google Research, where his mission is to build the theory and tools that make machine learning more reliable. After obtaining a PhD in computer science from Princeton University, Moritz spent three years at IBM Research Almaden prior to joining Google.