This area includes some presentations of my projects and personal lectures.
Simplicity does not precede complexity,
but follows it.
Alan Perlis
Nothing is obvious until somebody explains shortly How and Why something works.
Therefore, to be conformed to the KISS (Keep It Simple Stupid) design principle, we must implement a working deliverable first and then shaving away unnecessary assumptions or complexity in the problem. Once we know the available informations, the resources and what we want to achieve (in terms of time and space constraints), is possible to state the problem and identify a specific solution.
The presentations below were designed and delivered as 15 - 25 minute technical talks.
Therefore, to be conformed to the KISS (Keep It Simple Stupid) design principle, we must implement a working deliverable first and then shaving away unnecessary assumptions or complexity in the problem. Once we know the available informations, the resources and what we want to achieve (in terms of time and space constraints), is possible to state the problem and identify a specific solution.
The presentations below were designed and delivered as 15 - 25 minute technical talks.
- "Development of Tools for the Analysis of Geographic Variation and Classification of Geographic Peer Groups."
Given in May 2017 to the Wennberg International Collaborative Policy Meeting Conference in Melbourne: Better, Smarter Care - Reducing Unwarrented Variation. In this presentation, we introduce new methods and a tool to identify comparable reporting geographic areas. The tool is able to compare any two geographic areas and divide a set of geographies into groups with similar characteristics. The presentation can be downloaded here : Development of Tools for the Analysis of Geographic Variation and Classification of Geographic Peer Groups.
- " The conflict management in the Dempster and Shafer theory of evidence. "
Given in April 2014 to the University Commision of the Master in Automation and Management Engineering to introduce a single parameter model able to identify all the possible combination rules in the Smet's Transfer Belief Model.
Moreover, a new combination rule is proposed to converge towards a specific decision when the conflict is significative and an action must be taken. This rule is particularly suitable for real time applications. |
|
- " A New evidential distance measure based on the belief and plausibility assigment functions. "
Given in May 2014 to the Automation and Robotic University Lab Department to introduce the computational issues of the Fast Moebius Transform (FMT) - Data Fusion Algorithm and to propose a new metric for clustering and classification problem in order to filter the conflict among the system's agents.
However, this measure still have some problems for larger hypothesis set, hence a lot of work has to be done. At the moment I'm studying the topological properties of the probability space to investigate a possible connection between the combination rule and the convex polyhedra of the belief and plausibility functions. |
|