Text

  • Study location Ypsilon (U2-220), Campus Västerås
Date
  • 2024-12-03 09:45–15:30

Multidisciplinary Insights: Talks on AI, Stochastic Processes, and Statistical Methods

We invite you to attend "Multidisciplinary Insights: Talks on AI, Stochastic Processes, and Statistical Methods." This seminar will showcase leading-edge research and foster discussions on the applications of these fields. Join us to connect with experts, share perspectives, and look into insights from AI, stochastic processes, and statistics.

Date: December 3, 2024

Location: Ypsilon (U2-020), Malardalen University, Västerås

Organizer: Marko Dimitrov, marko.dimitrov@mdu.se

Sponsored by: Japan Society for the Promotion of Science (JSPS)

Registration includes lunch and fika, and the seminar is free of charge.

Featured Talk:

Title: MDP-based Decision Optimization: Applications to Maintenance Strategies for Systems Facing External Shocks

Speaker: Lu Jin, University of Electro-Communications, Tokyo

Abstract: Markov decision processes (MDPs) provide a powerful framework for decision optimization in complex systems. This talk illustrates the role of MDPs in guiding decision-making under uncertainty, with a focus on maintenance strategy optimization. As a case study, we consider systems subjected to both internal deterioration and external shocks, showcasing how MDPs can optimize condition-based maintenance policies to minimize costs.

We first analyze a single-unit system whose deterioration follows a discrete stochastic process, incorporating the interplay between internal deterioration and external shocks. This analysis is extended to a multi-unit system where each unit undergoes continuous stochastic deterioration. By formulating the maintenance decision problem as an MDP, we derive properties of the optimal policies that simplify their determination under reasonable assumptions. These results demonstrate how MDPs can structure and enhance decision-making processes in maintenance management.

Finally, numerical examples are provided to illustrate the application of the derived policies, highlighting the practical value of MDP-based optimization in reducing total expected discounted costs.

Schedule:

  • 09:45 – 10:15: Registering
  • 10:15 – 10:25: Introduction & JSPS presentation
  • 10:25 – 10:50: Jin Lu, MDP-based Decision Optimization: Applications to Maintenance Strategies for Systems Facing External Shocks
  • 10:50 – 11:15: Anna Friebe, Efficiently Bounding Deadline Miss Probabilities of Markov Chain Tasks
  • 11:15 – 11:40: Jean-Paul Murara, An Optimal Control Problem Involving Stochastic Behaviors
  • 11:40 – 12:05: Anatoliy Malyarenko, What is the spectral theory of random fields?
  • 12:05 – 13:00: Lunch
  • 13:00 – 13:25: Mara Kalicanin Dimitrov, Almost Exact Schemes for the Heston-type models
  • 13:25 – 13:50: Ayoub Haida, An Almost Exact Mixed Scheme for Double Mean Reverting Model
  • 13:50 – 14:15: Peter Backeman, Verifying ROS Systems Using Timed and Stochastic Timed Automata
  • 14:15 – 14:40: Cristina Seceleanu, Model-Checking and Reinforcement Learning for Autonomous Systems: Mission-Plan Synthesis and Collision Avoidance
  • 14:40 – 15:05: Dietrich von Rosen, Classification of growth curves
  • 15:05 – 15:30: Networking & Fika
  • 15:30: Closing

 

About JSPS External link.

The Japan Society for the Promotion of Science (JSPS) offers a variety of fellowship programs for overseas researchers, including short- and long-term opportunities to collaborate with Japanese institutions. Through programs like the “Postdoctoral Fellowships for Research in Japan” for young researchers and “Invitational Fellowships for Research in Japan” for mid-career and senior researchers, JSPS promotes international research collaboration and scientific advancement in Japan. These programs are ideal for researchers with an excellent track record seeking to expand their work while engaging with Japanese research communities.

We hope you will join us for this exciting event and take the opportunity to network with fellow professionals from a wide range of disciplines.

RSVP: Fill in this feedback form. External link.

Contact Information