6 papers
gLSTM: Mitigating Over-Squashing by Increasing Storage Capacity
Hugh Blayney, Álvaro Arroyo, Xiaowen Dong +1
Graph Neural Networks (GNNs) leverage the graph structure to transmit information between nodes, typically through the message-passing mechanism. While these models have found a wi…
On the Impact of Downstream Tasks on Sampling and Reconstructing Noisy Graph Signals
Baskaran Sripathmanathan, Xiaowen Dong, Michael Bronstein
We investigate graph signal reconstruction and sample selection for classification tasks. We present general theoretical characterisations of classification error applicable to mul…
Mathematical Foundations of Geometric Deep Learning
Haitz Sáez de Ocáriz Borde, Michael Bronstein
We review the key mathematical concepts necessary for studying Geometric Deep Learning.
Why do LLMs attend to the first token?
Federico Barbero, Álvaro Arroyo, Xiangming Gu +4
Large Language Models (LLMs) tend to attend heavily to the first token in the sequence -- creating a so-called attention sink. Many works have studied this phenomenon in detail, pr…
On Vanishing Gradients, Over-Smoothing, and Over-Squashing in GNNs: Bridging Recurrent and Graph Learning
Álvaro Arroyo, Alessio Gravina, Benjamin Gutteridge +5
Graph Neural Networks (GNNs) are models that leverage the graph structure to transmit information between nodes, typically through the message-passing operation. While widely succe…
Enhancing the Expressivity of Temporal Graph Networks through Source-Target Identification
Benedict Aaron Tjandra, Federico Barbero, Michael Bronstein
Despite the successful application of Temporal Graph Networks (TGNs) for tasks such as dynamic node classification and link prediction, they still perform poorly on the task of dyn…