1: \begin{abstract}

2: Many real-world phenomena can be modeled as a

3: graph, making them extremely valuable due to their ubiquitous

4: presence. GNNs excel at capturing those relationships and pat-

5: terns within these graphs, enabling effective learning and prediction tasks. GNNs are constructed using Multi-Layer Perceptrons

6: (MLPs) and incorporate additional layers for message passing

7: to facilitate the flow of features among nodes. It is commonly

8: believed that the generalizing power of GNNs is attributed to

9: the message-passing mechanism between layers, where nodes

10: exchange information with their neighbors, enabling them to

11: effectively capture and propagate information across the nodes of

12: a graph. Our technique builds on these results, modifying

13: the message-passing mechanism further: one by

14: weighing the messages before accumulating at each node and

15: another by adding Residual connections. These two mechanisms

16: show significant improvements in learning and faster

17: convergence.

18: 

19: 

20: 

21: \end{abstract}

22: