Events

Title:
Metaheuristics - the metaphor exposed

Abstract:
The widespread introduction of metaphor-based metaheuristics has led to a surge of algorithms whose novelty and scientific grounding are often questionable. Many so-called innovations are rebranded versions of existing methods, while the metaphors themselves are frequently inconsistent with the algorithmic principles they aim to represent. In this presentation, I critically analyze specific cases where the lack of originality and flawed metaphors have been exposed. Additionally, I explore actions taken by the research community to address this proliferation, including efforts to promote more rigorous standards for algorithm design, publication, and evaluation. By combining critique with a discussion of mitigation strategies, this talk aims to foster a more principled and scientifically robust approach to metaheuristic research, moving beyond metaphorical allure toward genuine contributions to the field.

Short Bio:
Kenneth Sörensen obtained his PhD entitled "A framework for robust and flexible optimization using metaheuristics" from the University of Antwerp in 2003 and continued as a post-doctoral researcher at the UAntwerp. From 2006 to 2009, he worked as a postdoc researcher of the Flemish Fund for Scientific Research (FWO) at the Faculty of Engineering of the KULeuven. In 2009, Kenneth Sörensen was appointed as a Research Professor of the Faculty of Business and Economics of the University of Antwerp, a position he holds to date. Within this Faculty, he founded the ANT/OR research group, which focuses on applications of operations research and currently consists of three professors and about ten PhD students. Under his supervision, over twenty PhD's were obtained. Kenneth Sörensen has published a large number of articles in international refereed journals and has presented his work at numerous scientific conferences. He is considered one of the world’s leading experts in the field of metaheuristics. His main research interest lies in the application of advanced (metaheuristic) optimization methods and in the development and study of optimization methods. Kenneth Sörensen is the founder and current coordinator of EU/ME – the EURO working group on metaheuristics, the largest online platform for researchers in metaheuristics worldwide. More information on Kenneth Sörensen's research, teaching, and other activities can be found on his personal website at the University of Antwerp.

https://oficinavirtual.ugr.es/redes/SOR/SALVEUGR/accesosala.jsp?IDSALA=22996529
- Contraseña de la reunión: 481133

Title: Generative models as queryable world models

Abstract:  Over the last decade, the de facto standard for training high performing representation learning models has heavily relied on large scale static datasets crawled from the Internet. However, recent advances in visual content creation are challenging this status quo by pushing researchers to leverage high performing image generative models as world models that work in tandem with representation learning models by providing them with data. In this talk, I will address two research questions: (1) Are state-of-the-art image generative models optimized to work as world models?; and (2) What is the most effective way to guide the generative model to produce samples that are useful for training downstream models?

 
Bio: Adriana Romero-Soriano is currently a research scientist at Meta AI (FAIR), an adjunct professor at McGill University, a core industry member of Mila, and the recipient of a Canada CIFAR Chair. The playground of her research has been defined by problems which require inferring full observations from limited sensory data, building models of the world with the goal to improve impactful downstream applications responsibly. Her most recent research focuses on improving the quality, consistency, and representation diversity of visual content creation systems. Adriana received her Ph.D. from University of Barcelona, where she worked with Dr. Carlo Gatta, and spent two years as post-doctoral researcher at Mila working with Prof. Yoshua Bengio.

Para entrar en la reunión, pulse el siguiente enlace:
https://oficinavirtual.ugr.es/redes/SOR/SALVEUGR/accesosala.jsp?IDSALA=22996576
- Contraseña de la reunión: 249099