This paper introduces a hybrid model combining quantum feature encoding with neural attention to enhance robustness in real-time visual analytics. The system improves feature localization, reduces noise sensitivity, and delivers stable performance across varying conditions in high-resolution data environments.

Amit Rajan Desai, Jean Marc Laurent, Maria Cristina Navarro, Rakesh Pratap Iyer, Elena Sophie Rossi

This research presents a quantum graph neural network framework for node classification in temporal-spatial networks. It enhances interpretability using entangled feature embeddings, ensuring accurate and explainable predictions across dynamic, sensor-rich datasets with complex spatial interdependencies and evolving connectivity.

Liu Wei Han, Zhang Yu Chen, Gao Ming Rui, Xu Lin Bo, Chen Zhen Liang, Huang Xiang Wen

A federated continual learning model is introduced with modules that retain task knowledge over time. It adapts to new data streams without compromising older learning, maintaining accuracy while adhering to strict privacy requirements in decentralized, edge-based AI applications.

Neha Meenakshi Sharma, Taro Kazuki Watanabe, Carlos Diego Morales, Pooja Lakshmi Iyer, Jean Francois Renault

This work proposes an adaptive feature selection technique that uses mutual information optimization to identify key signals in genomic datasets. The model significantly reduces dimensionality and enhances classification performance in health informatics applications involving complex biomedical data structures.

Emma Charlotte Douglas, Laura Jane Whittaker, Thomas Benjamin Clarke, Nathan William Frost, Olivia Isabella Greene

This study introduces a multi-granular attention model for extracting temporal relations in clinical text. The model leverages distant supervision, temporal embedding alignment, and hierarchical context windows to improve accuracy in time-sensitive event linking across medical narratives.

Chen Yu Bo, Xu Hao Liang, Zhang Ming Jie, Liu Fang Lei, Gao Xiu Tian, Wang Qing Zhao

We propose a deep fusion model that enhances image super-resolution using residual transfer and spatial attention across resolution layers. The model outperforms traditional upscaling methods by retaining fine-grained textures and improving perceptual consistency across degraded and restored image spaces.

David Edward Simmons, Rachel Anne Johnson, Michael Louis Patterson, Sarah Abigail Grant, Emily Victoria Adams

We propose a transformer architecture that aligns semantic embeddings across audio and visual inputs. It enhances scene interpretation in multimodal environments by fusing contextual representations and temporal signals, achieving improved performance on cross-modal understanding and classification benchmarks.

Michael James Harrington, Sarah Elizabeth Collins, Olivia Katherine Moore, Robert Thomas Adams, Emily Lauren Carter