A series of meetings designed to allow post-doc and Ph.D. students to present their research topics and promote collaborations.
On one Wednesday each month, a different research topic is presented in simple and accessible terms.
The seminars are designed to be attended by post-doc and Ph.D. students, but graduates and undergraduates are welcome.
Ph.D. courses in mathematics are coordinated by Prof. Alessandro Giuliani.
Enroll yourself or visualise the events calendar.
Abstract. Single Image Super Resolution (SISR) is an ill-posed problem: the lack of an intrinsic notion of “same content” across resolutions motivates its statistical formulation. We begin by clarifying this viewpoint and its relevance to our medical imaging setting. We then present an approach based on the additive refinement of the wavelet coefficients of a classically upsampled image. Building on this formulation, we introduce a super-resolution architecture that implements it by combining two-dimensional discrete wavelet transforms with the Swin Transformer. We conclude by discussing the training strategy and the use of knowledge distillation, along with quantitative and qualitative results on medical datasets.
Abstract. Inspired by recent works exploring the interconnections between formal knot theory and quiver representation theory, we introduce new combinatorial objects generalizing the notion of a Kauffman state. A BMS state consists of a graph \(G\) embedded in an orientable surface, endowed with a nonnegative weight on the edges and a pair of integer-valued functions on the set of angles satisfying a balancing condition. Such an object is naturally associated with a representation of the Jacobian algebra of the medial quiver of \(G\). After defining two symmetric notions of partial order for BMS states, we study the structure of the category in which these objects live, in connection with the lattices of subrepresentations arising from them. Furthermore, we address questions of decomposability. Finally, we discuss polynomial aspects and outline some open problems currently under investigation.
Partecipants.
Partecipants.
Partecipants.
Partecipants.