Aaltodoc - homepage
Communities & Collections
Browse Aaltodoc publication archive
EN | FI |
Log In
  1. Home
  2. Browse by Author

Browsing by Author "Aunula, Jerry"

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • Results Per Page
  • Sort Options
  • No Thumbnail Available
    An Algorithm for Generating Most Probable Paths in Decision Programming
    (2021-10-01) Aunula, Jerry
    Perustieteiden korkeakoulu | Bachelor's thesis
  • Loading...
    Thumbnail Image
    Models for Solving Multi-Agent Decision Problems
    (2023-12-12) Aunula, Jerry
    Perustieteiden korkeakoulu | Master's thesis
    Several ways of solving multi-agent decision problems, e.g., complete information game theory and adversarial risk analysis, have been proposed in the literature. Most approaches, however, cannot accommodate constraints spanning the entire problem or situations in which earlier decisions cannot be recalled when making later ones. Decision Programming is a methodology that helps address these limitations; however, it has thus far only been used to solve single-agent decision problems. This thesis aims to expand Decision Programming to accommodate multi-agent decision problems. Our approach first transforms the multi-agent problem into several single-agent problems, one for each actor, that are represented with an influence diagram. We solve these problems with the level-$k$ approach, updating the initial distribution for the decisions of each actor iteratively. For each problem at each level, we use Decision Programming to solve the single-agent problems. This process is repeated until we arrive at a convergent solution. We apply our methodology to two examples, one in critical infrastructure protection and the other in border security. In the first example, our methodology produced results very similar to a traditional adversarial risk analysis method that used node removals. Furthermore, we solved an extension of the first example and one example concerning border security, which both included overarching constraints, non-perfect strategists, and multiple decision alternatives. Overall, our solutions created beneficial results for protecting critical infrastructure and constructing border security portfolios.
Help | Open Access publishing | Instructions to convert a file to PDF/A | Errata instructions | Send Feedback
Aalto UniversityPrivacy notice | Cookie settings | Accessibility Statement | Aalto University Learning Centre