Untrained graph neural networks for denoising

S Rey, S Segarra, R Heckel… - IEEE Transactions on …, 2022‏ - ieeexplore.ieee.org
A fundamental problem in signal processing is to denoise a signal. While there are many
well-performing methods for denoising signals defined on regular domains, including …

Convolutional learning on directed acyclic graphs

S Rey, H Ajorlou, G Mateos - arxiv preprint arxiv:2405.03056, 2024‏ - arxiv.org
We develop a novel convolutional architecture tailored for learning from data defined over
directed acyclic graphs (DAGs). DAGs can be used to model causal relationships among …

A robust alternative for graph convolutional neural networks via graph neighborhood filters

VM Tenorio, S Rey, F Gama, S Segarra… - 2021 55th Asilomar …, 2021‏ - ieeexplore.ieee.org
Graph convolutional neural networks (GCNNs) are popular deep learning architectures that,
upon replacing regular convolutions with graph filters (GFs), generalize CNNs to irregular …

Robust graph-filter identification with graph denoising regularization

S Rey, AG Marques - ICASSP 2021-2021 IEEE International …, 2021‏ - ieeexplore.ieee.org
When approaching graph signal processing tasks, graphs are usually assumed to be
perfectly known. However, in many practical applications, the observed (inferred) network is …

Overparametrized deep encoder-decoder schemes for inputs and outputs defined over graphs

S Rey, V Tenorio, S Rozada, L Martino… - 2020 28th European …, 2021‏ - ieeexplore.ieee.org
There is a growing interest in the joint application of graph signal processing and neural
networks (NNs) for learning problems involving complex, non-linear and/or non-Euclidean …

On architecture selection for linear inverse problems with untrained neural networks

Y Sun, H Zhao, J Scarlett - Entropy, 2021‏ - mdpi.com
In recent years, neural network based image priors have been shown to be highly effective
for linear inverse problems, often significantly outperforming conventional methods that are …

Deep encoder-decoder neural network architectures for graph output signals

S Rey, V Tenorio, S Rozada, L Martino… - 2019 53rd Asilomar …, 2019‏ - ieeexplore.ieee.org
Neural networks (NNs) and graph signal processing have emerged as important actors in
data-science applications dealing with complex (non-linear, non-Euclidean) datasets. In this …

Generative adversarial networks for graph data imputation from signed observations

A Madapu, S Segarra, SP Chepuri… - ICASSP 2020-2020 …, 2020‏ - ieeexplore.ieee.org
We study the problem of missing data imputation for graph signals from signed one-bit
quantized observations. More precisely, we consider that the true graph data is drawn from a …

Neighborhood Graph Neural Networks under Random Perturbations and Quantization Errors

LB Saad, AN Nama… - 2022 IEEE 23rd …, 2022‏ - ieeexplore.ieee.org
Graph convolutional neural networks (GCNNs) have emerged as a promising tool in the
deep learning community to learn complex hidden relationships of data generated from non …