The field of graph neural networks (GNNs) is rapidly evolving, with a focus on improving performance, scalability, and interpretability. Recent developments have led to the creation of novel architectures, such as sheaf graph neural networks and graph agentic networks, which address limitations in existing models. These advancements enable more effective learning on heterogeneous and dynamic graphs, with applications in node classification, graph anomaly detection, and stochastic dynamics modeling. Furthermore, the integration of GNNs with other techniques, like contrastive learning and optimal transport, has shown promising results. Noteworthy papers in this area include those that propose innovative solutions for graph edit distance, online continual graph learning, and hierarchical contrastive learning on text-attributed hypergraphs. Additionally, research in related fields, such as crowd dynamics and process modeling, has leveraged machine learning and graph-based methods to improve simulation and control. Overall, the field is moving towards more robust, flexible, and interpretable models that can handle complex graph-structured data. Notable papers include: Sheaf Graph Neural Networks via PAC-Bayes Spectral Optimization, which introduces a novel scheme for robust semi-supervised node classification. WeightFlow: Learning Stochastic Dynamics via Evolving Weight of Neural Network, which presents a paradigm for modeling dynamics directly in the weight space of a neural network.