Algorithmic stability for regression and classification

Add to your list(s) Download to your calendar using vCal

• Rina Foygel Barber (Chicago)
• Thursday 13 March 2025, 16:00-17:00
• Centre for Mathematical Sciences MR2.

If you have a question about this talk, please contact HoD Secretary, DPMMS.

In a supervised learning setting, a model fitting algorithm is unstable if small perturbations to the input (the training data) can often lead to large perturbations in the output (say, predictions returned by the fitted model). Algorithmic stability is a desirable property with many important implications such as generalization and robustness, but testing the stability property empirically is known to be impossible in the setting of complex black-box models. In this work, we establish that bagging any black-box regression algorithm automatically ensures that stability holds, with no assumptions on the algorithm or the data. Furthermore, we construct a new framework for defining stability in the context of classification, and show that using bagging to estimate our uncertainty about the output label will again allow stability guarantees for any black-box model. This work is joint with Jake Soloff and Rebecca Willett.

A wine reception in the Central Core will follow this lecture

This talk is part of the Peter Whittle Lecture series.

This talk is included in these lists:

• All CMS events
• All Talks (aka the CURE list)
• CMS Events
• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• Cambridge talks
• Centre for Mathematical Sciences MR2
• Chris Davis' list
• DAMTP Departmental Seminar
• DAMTP info aggregator
• DPMMS Lists
• DPMMS info aggregator
• DPMMS lists
• Faculty of Mathematics Lectures
• Guy Emerson's list
• Hanchen DaDaDash
• Interested Talks
• Machine Learning
• Peter Whittle Lecture
• Probability
• School of Physical Sciences
• Statistical Laboratory info aggregator
• Statistics
• Statistics Group
• bld31
• custom
• ndb35's list
• rp587

Note that ex-directory lists are not shown.