Statistics seminar: Consistent Sequential Learning Algorithms for Highly Dependent Time Series

3 March 2017, 2.15 PM - 3 March 2017, 3.15 PM

Azadeh Khaleghi, Lecturer in Statistical Learning, Lancaster University

SM3, School of Mathematics

Azadeh Khaleghi, Lecturer in Statistical Learning, Lancaster University

Title:  Consistent Sequential Learning Algorithms for Highly Dependent Time Series
Abstract: One of the main challenges in statistical learning today is to make sense of complex sequential data which
typically represent interesting, unknown phenomena to be inferred. To address this problem from a mathematical perspective, it is usually assumed that data have been generated by some random process where the goal is to make inference about the stochastic mechanisms that produce the samples. The typical assumptions that samples are generated i.i.d or that their distributions belong to specific model classes (e.g. HMMs) can, at times, undermine the possibly complex structure of the data and the potentially long-range dependencies. Moreover, since little is usually known about the nature of the data, it is important to address inference beyond parametric and modelling assumptions. One approach is to assume that the process distributions are stationary ergodic but do not belong to any simpler class of processes. This paradigm has proved useful in a number of learning problems that involve dependent sequential data. At the same time, many natural problems already turn out to be impossible to solve under this assumption. In this talk I will discuss the possibilities and limitations of sequential inference in the stationary ergodic framework.


Contact information

Organisers: Haeran Cho

Edit this page