Of possible interest for someone in the SIPTA community 🙂

Yours,

Michele

Da: Paolo Perrone <paolo.perrone@cs.ox.ac.uk> Inviato: lunedì 14 ottobre 2024 10:08 A: talk-announce@cs.ox.ac.uk <talk-announce@cs.ox.ac.uk>; Rob Cornish <rob.cornish@stats.ox.ac.uk>; Kaushik Chintam <k.chintam@se23.qmul.ac.uk>; Michele Caprio <michele.caprio@manchester.ac.uk> Oggetto: OASIS talk next Friday: Michele Caprio - Imprecise Probability in Machine Learning

Hi all,

The OASIS seminar is back! (https://www.cs.ox.ac.uk/seminars/oasis/ [cs.ox.ac.uk]<https://urldefense.com/v3/__https://www.cs.ox.ac.uk/seminars/oasis/__;!!PDiH...>)

This Friday at 2 PM AM in Lecture Theatre A we are going to have an OASIS talk by Michele Caprio (University of Manchester), titled "Imprecise Probabilistic Machine Learning: Being Precise About Imprecision". The abstract is below.

Before the talk some of us are planning to have lunch with the speaker. If you are interested in joining us, please let me know by Thursday morning.

The talk is also accessible remotely, at the following Zoom link: https://cs-ox-ac-uk.zoom.us/j/99562853810?pwd=VmJFUlhNbVJwOUpjTE5hUk9LUkMvUT... [cs-ox-ac-uk.zoom.us]<https://urldefense.com/v3/__https://cs-ox-ac-uk.zoom.us/j/99562853810?pwd=Vm...> Meeting ID: 995 6285 3810 Passcode: 983333

For further information see the following links, https://www.cs.ox.ac.uk/seminars/oasis/ [cs.ox.ac.uk]<https://urldefense.com/v3/__https://www.cs.ox.ac.uk/seminars/oasis/__;!!PDiH...> https://www.cs.ox.ac.uk/seminars/2688.html [cs.ox.ac.uk]<https://urldefense.com/v3/__https://www.cs.ox.ac.uk/seminars/2653.html__;!!P...>

All the best, Paolo

Speaker: Michele Caprio (University of Manchester) Title: Imprecise Probabilistic Machine Learning: Being Precise About Imprecision

Abstract: This talk is divided into two parts. I will first introduce the field of “Imprecise Probabilistic Machine Learning”, from its inception to modern-day research and open problems, including motivations and clarifying examples. In the second part, I will present some recent results that I've derived on Imprecise Markov Processes. I will introduce the concept of an imprecise Markov semigroup Q. It is a tool that allows to represent ambiguity around both the initial and the transition probabilities of a Markov process via a compact collection of plausible Markov semigroups, each associated with a (different, plausible) Markov process. I will use techniques from geometry, functional analysis, and (high dimensional) probability to study the ergodic behavior of Q. I will show that, if the initial distribution of the Markov processes associated with the elements of Q is known and invariant, under some conditions that also involve the geometry of the state space, eventually the ambiguity around their transition probability fades. I call this property ergodicity of the imprecise Markov semigroup, and I will relate it to the classical notion of ergodicity. I will present ergodicity when the state space is Euclidean or a Riemannian manifold. The importance of my findings for the fields of machine learning and computer vision will also be discussed.

This talk is based on the following work, https://arxiv.org/abs/2405.00081 [arxiv.org]<https://urldefense.com/v3/__https://arxiv.org/abs/2405.00081__;!!PDiH4ENfjr2...>, where I also extend the results to the case where the state space is an arbitrary measurable space.