Extra Colloquium Mathematics - Max Welling
Join us for coffee and tea at 15.45 p.m.
Date: Thursday, March 17th 2011
Speaker: Max Welling
Room: 5161.0289 (Bernoulliborg),
Time: 16.15
Title: Herding: Learning with Weakly Chaotic Nonlinear Dynamical Systems
Abstract:
We describe a class of deterministic weakly chaotic dynamical systems with infinite memory. These ``herding systems'' combine learning and inference into one algorithm. They convert moments directly into a sequence of pseudo-samples without learning an explicit model. Using the "perceptron cycling theorem" we can easily show that Monte Carlo estimates based on these pseudo-samples converge at an optimal rate of O(1/T), due to infinite range negative auto-correlations. We show that the information content of these sequences, as measured by sub-extensive entropy, can grow as fast as K log(N), which is faster than the usual 1/2 K log(N) for sequences generated by random posterior sampling from a Bayesian posterior model. In continuous spaces we can control an infinite number of moments
by formulating herding in a Hilbert space. Also in this case sample averages over arbitrary functions in the Hilbert space will converge at an optimal rate of O(1/T).
More generally, we advocate the application of the rich theoretical framework of nonlinear dynamical systems and chaos theory to statistical learning.
Colloquium coordinators are Prof.dr. A.C.D. van Enter (e-mail : A.C.D.van.Enter@rug.nl) and
Dr. M. Dür (e-mail: M.E.Dur@rug.nl)
| Last modified: | 10 February 2021 2.28 p.m. |
More news
-
08 October 2025
Not all plastic needs to be bio-based or biodegradable
Per person, we throw away about 33 kilos of plastic packaging per year. Professor of Polymer Chemistry Katja Loos is working on a more sustainable future for plastics - by looking at more than the material itself.
-
06 October 2025
The GenAI-bubble will burst, but don’t give up on AI altogether
'People keep promoting the belief that generative AI provides universal tools that are capable of much more,’ says Michael Biehl, Professor of Machine Learning. ‘Sooner or later, the genAI bubble will burst,’ he is certain. But that doesn’t mean all...
-
01 October 2025
In Science Podcast: Ajay Kottapalli about seal whiskers and ultrasensitive sensors
'In Science' is the podcast of the University of Groningen. In this episode, we’re joined by Ajay Kottapalli, Associate Professor at the Engineering and Technology Institute Groningen and co-founder of the Sencilia startup.