Skip to content

Image title

Giovanni Misitano, PhD

We have all kinds of fancy AI tools that can predict this and that, but these tools are complete black-boxes from the perspective of most decision makers, and humans general. When it comes to making actual decisions based on the output of these tools, we lack decision-support and grounds to justify our solutions. My research can help us make informed and explicable decisions when multiple conflicting criteria are to be considered without a clear 'best' solution. I am all about providing support and tools to decision makers (domain experts) to make the best possible decisions given their expertise, preferences, and the available data and information. I do not care much about replacing human decision makers, but I do think we can do a much better job at supporting decision makers across a plethora of different domains, without having to rely on complete black-boxes.

Stop guessing and start optimizing—and do it explainably!

About me

I am currently employed as a post-doctoral researcher at the University of Jyväskylä, Finland, in the Multiobjectove Optimization Group led by Professor Kaisa Miettinen. I specialize in methods and tools for multiobjective optimization. These methods can be applied anywhere where data is available, there are conflicting objectives to be optimized, and there are decision makers with relevant domain expertise. That is, pretty much anywhere these days! My current work focuses on researching decision support tools, more specifically, interactive multiobjective optimization methods, that are explainable and transparent to decision makers and stakeholders alike. I also have a deep interest in developing user interfaces, and the user experience in general from a cognitive perspective.

I am also a main contributor and maintainer of DESDEO. DESDEO is an open-source Python framework for interactive multiobjective optimization. Our research group is currently overhauling the framework, working towards a ''DESDEO 2.0.'' This version will be a full-stack solution, with a backend containing the algorithmic implementations of various interactive multiobjective optimization methods, both multi-criteria decision-making methods and stochastic evolutionary methods; a web application programming interface (API) with database and user session management to connect the backend over the web to existing and custom interfaces; and a frontend, a graphical web-based interface designed by us. This is a huge endeavor, but it is fueled by our research group's decades of collective experience in researching and developing multiobjective optimization methods.

In the future, DESDEO will offer huge opportunities in academia and industry for tackling data-based problems with multiple and conflicting objectives functions to be either minimized or maximized simultaneously. DESDEO can very much function as the connecting piece between all the fancy AI tools available these days, and the act of making actual decisions based on these tools' output.

Check out the development of the new version of DESDEO here. In the documentation, there is also a primer on what multiobjective optimization is, check the primer here.

Note

Everything under the desdeo2 branch is under active development and subject to sudden, and possibly drastic, changes. We will release a more stable version going into the latter half of the year 2024.

Contents

  • Research Experience

    Check my past and current research experience and projects I am, and have been, involved in.

    Research Experience

  • Publications

    Read my publications, which are all either open access, or have been published in parallel on open access platforms!

    Publications

  • Activities and Rewards

    Browse my current and past positions, rewards, nominations, and grants.

    Activities and Rewards

  • Education and Teaching

    See the degrees I have earned, and my teaching experience at a university level.

    Education and Teaching

  • Dissemination

    Have a look at my conference presentations, invited talks and lectures, posters, and more. Slides and recordings are available for many of my past dissemination activities!

    Dissemination

  • Get a homepage like this!

    This page has been built using the materials theme for mkdocs. The source code of this page is completely open and very easy to modify and update.

    Visit GitHub