Publications
Peer-reviewed work and talks
GNNs for Time Series Anomaly Detection: An Open-Source Framework and a Critical Evaluation
AuthorWe present GraGOD, an open-source framework for time series anomaly detection (TSAD) using Graph Neural Networks, designed to support reproducible experimentation across datasets, graph structures, and evaluation strategies. Our critical evaluation shows that GNNs improve both detection performance and interpretability, and that attention-based GNNs remain robust when the graph structure is uncertain or inferred.
Federico Bello, Gonzalo Chiarlone, Marcelo Fiori, Gastón García González, Federico Larroca
The Lattice-Input Discrete-Time Poisson Channel
AuthorIn this work we examine the Lattice-Input Discrete-Time Poisson (LIDTP) channel, relevant to DNA storage systems. We analyze communication performance with lattice-constrained inputs under Poisson noise and discuss implications for reliable molecular data storage.
Federico Bello, Álvaro Martín, Tatiana Rischewski, Gadiel Seroussi
Conferences
- LANET 2025Aug 2025Punta del Este, Uruguay
GraGOD: Graph based time series anomaly detection
SpeakerIn this conference I presented my thesis work, where we studied modeling multivariate time series as graphs and using GNNs to exploit inter-variable structure for reliable anomaly detection.