28-29 November 2024
Since the success of hidden Markov models (HMM) in the last decades of the 20th century, HMM-related modeling has been rapidly evolving. Often HMMs is a too limited model class, thus there is an increasing interest towards more general, flexible and complex models where the HMM-like dynamic programming algorithms are still applicable. One such surprisingly rich class of stochastic models is the so-called multiple (multidimensional) Markov models, including pairwise and triplet Markov models. A marginal or observed process of such a multiple model might lack the Markov property itself (allowing modeling the long dependence), but the joint Markov property makes dynamic programming algorithms often work.
The Tartu Workshop on Markov Modeling (28-29 November 2024, Tartu, Estonia), will bring together experts in different areas of Markov modeling. The topics of the workshop cover both theory and applications: Markov models, hidden (semi)-Markov models, asymptotics, dynamic programming, applications of Markov models in genetics, statistical inference, parameter estimation, ergodic theory, segmentation, evolution models, applications of Markov models in genetics, Bayesian approach, sequence comparison.
Programme
Workshop Abstracts you can find here:
Thursday, November 28 (room 1008)
- 9:00-9:45 Paul Doukhan (CY Cergy Paris University) Concentration inequalities for some infinite memory models
- 9:45-10:30 Evgeny Verbitskiy (Leiden University/University of Groningen) Hidden Markov processes and fields
- 11:00-11:45 Chris Watkins (Royal Holloway, University of London) Evolutionary models for sets and networks, that satisfy detailed balance
- 11:45-12:30 Marc Escudier (Telecom Sudparis) On the robustness of pairwise Kalman predictors
- 12:30-13:00 Oskar Soop (University of Tartu) No Viterbi for triplet Markov models — a method to find the most likely sequence of states of a marginal process
- 14:00-14:45 David Bolin (King Abdullah University of Science and Technology) Markov properties of Gaussian processes on metric graphs
- 14:45-15:30 Alexey Koloydenko (Royal Holloway, University of London) Can Markov modelling help advance the novel Autofluorescence-Raman technology to diagnose and treat skin and breast cancers?
- 16:00-16:45 Asger Hobolth (Aarhus University) Posterior analyses and decoding of hidden Markov models with applications to identification and characterization of archaic fragments in human genomes
- 16:45-17:30 Daniela Bertacchi (University of Milano-Bicocca) A new stochastic order applied to branching random walks
Friday, November 29 (room 1006)
- 9:30-10:15 Sangita Kulathinal and Dario Gasbarra (University of Helsinki) Non-homogeneous and hidden Markov multistate models for intermittently observed processes: application to partially observed treatment outcomes among patients with nAMD
- 10:15-11:00 Aapeli Nevala (University of Helsinki) Continuous-time hidden Markov model for colorectal cancer – approaches to parameter estimation and common problems
- 11:30-12:15 Etienne Sebag (University of Helsinki) Hidden Markov models for longitudinal data in multistate models: comparing algorithms for parameter and hidden state estimation
- 12:15-13:00 Joonas Sova (University of Tartu) Behaviour of extremal optimal alignments for Markov chains
How to get to Tartu
Tartu is situated in Southern Estonia and has a direct flight connection to Helsinki, but you can also travel to Tartu through Tallinn (approx 180 km from Tartu) and Riga (approx 250 km from Tartu).
Tartu Airport: Helsinki-Tartu flights are operated by Finnair. The airport is located about 10 km from the city centre, and you can use a taxi or shuttle to get to the city. For more information about flights to Tartu and airport shuttle click here.