Institution profile

University of Coimbra

Academic institutioneurope · pt
Official website
Research library98linked papers
Opportunities0open roles
Selected work

Representative Papers

Free Doubly-Infinitary Distributive Categories are Cartesian Closed

Mar 15, 2024arXiv.org

This paper investigates *bifinitely distributive categories*—categories in which products distribute over coproducts and coproducts distribute over products—in the infinitary setting, clarifying their logical relationships with universality, infinitary distributivity, and Cartesian closure. Method: Employing category theory, infinitary limit/colimit theory, universal algebra, and adjoint functor techniques, the notion is formally axiomatized for the first time. Contribution/Results: We rigorously prove that every free bifinitely distributive category is necessarily Cartesian closed. Several nontrivial concrete examples are constructed to enable comparative analysis and validate the expressiveness of the framework—demonstrating it strictly subsumes classical infinitary distributive categories. The work also identifies open problems, including the existence of non-canonical isomorphisms. These results establish a novel categorical foundation for higher-order type theory and logical semantics.

2 citationsRead paper

Trigger Optimization and Event Classification for Dark Matter Searches in the CYGNO Experiment Using Machine Learning

Jan 28, 2026

This work addresses the challenges of efficient real-time triggering, compression, and background suppression in high-resolution sparse optical imaging within the CYGNO experiment. To this end, two novel approaches are proposed: first, an unsupervised online compression framework based on a convolutional autoencoder that extracts regions of interest (ROIs) via reconstruction residuals, achieving fully unsupervised real-time ROI identification for the first time in CYGNO—retaining 93.0% of signal intensity while discarding 97.8% of background pixels with only 25 ms inference latency; second, the application of the weakly supervised Classification Without Labels (CWoLa) method to mixed data, which successfully identifies compact, circular nuclear recoil events and achieves performance approaching the theoretical optimum.

1 citationsRead paper
Recent publications

Latest Papers

Autonomous Telerehabilitation via Skeletal Motion Prediction and Joint-Level Performance Assessment

Aug 12, 2026

This work proposes a markerless RGB video–based dual-module system to enable autonomous, therapist-free remote rehabilitation by simultaneously performing action recognition and generating structured feedback. The method innovatively integrates a self-attention bidirectional LSTM with graph-structured motion prediction (STARS) to achieve end-to-end joint-level action quality assessment and short-term pose forecasting. Feedback signals are refined through MMD-NCA metric learning and MPJPE error optimization. Experimental results demonstrate that the system achieves an average class accuracy of 96.45% for action classification on the PROZIS dataset. Furthermore, STARS attains an MPJPE of 75.8 mm over a 560 ms prediction horizon on Human3.6M, outperforming existing baselines.

0 citationsRead paper