With the rapid advancements in artificial intelligence (AI) and mobile communication networks, particularly the advent of 5G and 6G, there is a growing need for efficient distributed machine learning methods. Federated Learning (FL), which enables distributed training while maintaining data privacy, has emerged as a promising solution. However, the communication efficiency in FL remains a significant challenge, especially as the model size and number of participants increase. Despite the progress in both AI and mobile communications, existing research tends to focus on one domain at a time, either mobile communication or AI, without fully leveraging the critical features of both fields. In this paper, we propose a training-oriented data chunking and transmission method designed to parallelize data transmission and model updates. We further introduce a dynamic precision adaptation method, optimizing communication efficiency. Through our proposed approach, we not only reduce communication overhead but also enhance model training performance, enabling more efficient and scalable distributed learning in the context of 5G and 6G networks. The method offers a promising solution to the communication bottleneck in Federated Learning, paving the way for more effective AI applications in next-generation mobile networks.

Realizing Generalized Zero-Shot Learning (GZSL) based on large models is emerging as a prevailing trend. However, most existing methods merely regard large models as black boxes, solely leveraging the features output by the final layer while disregarding potential performance enhancements from other layers. Indeed, numerous researchers have visually depicted variations in the features learned across different layers of neural networks. Motivated by this observation, we propose a Vision Transformer (ViT)-based GZSL method named Depth-Aware Multi-Modal ViT (DAM2ViT), which exploits multi-level features of ViT. DAM2ViT incorporates a multi-modal interaction block to align semantic information of categories across multiple layers, thereby augmenting the model's capacity to learn associations between visual and semantic spaces. Extensive experiments conducted on three benchmark datasets (i.e., CUB, SUN, AWA2) have showcased that DAM2ViT achieves competitive results compared to state-of-the-art methods.

The EAE task extracts a structured event record from an event text.Most existing approaches train the EAE model on each dataset independently and ignore the overlap knowledge across datasets.However, insufficient event records in a single dataset often prevent the existing model from achieving better performance.In this paper, we clearly define the overlap knowledge across datasets and split the knowledge of the EAE task into overlap knowledge across datasets and specific knowledge of the target dataset.We propose APE model to learn the two parts of knowledge in two serial learning phases without causing catastrophic forgetting.In addition, we formulate both learning phases as conditional generation tasks and design Stressing Entity Type Prompt to close the gap between the two phases.The experiments show APE achieves new state-of-the-art with a large margin in the EAE task.When only ten records are available in the target dataset, our model dramatically outperforms the baseline model with average 27.27%F1 gain.