The increasing utilization of newly emerging networks (e.g., private-5G) across industries underscores the need for accurate traffic prediction to manage network resources effectively. However, rapidly emerging networks face challenges in accurate prediction due to limited training data at the early stage and fluctuation in traffic load at the maintenance stage. In response, we propose ISOML (Inter-Service Online Meta-Learning), a novel traffic prediction pipeline designed for newly emerging networks. ISOML utilizes meta-learning to address data scarcity and employs the EWC (Elastic Weight Consolidation) for online learning to learn dynamics of traffic patterns. Experimental validation in real-world datasets demonstrates the efficacy of ISOML in predicting traffic for emerging network environments.
The emergence of hyper-realistic deepfake videos has raised significant concerns regarding their potential misuse. However, prior research on deepfake detection has primarily focused on image-based approaches, with little emphasis on video. With the advancement of generation techniques enabling intricate and dynamic manipulation of entire faces as well as specific facial components in a video sequence, capturing dynamic changes in both global and local facial features becomes crucial in detecting deepfake videos. This paper proposes a novel sequential attentive face embedding, SAFE, that can capture facial dynamics in a deepfake video. The proposed SAFE can effectively integrate global and local dynamics of facial features revealed in a video sequence using contrastive learning. Through a comprehensive comparison with the state-of-the-art methods on the DFDC (Deepfake Detection Challenge) dataset and the FaceForensic++ benchmark, we show that our model achieves the highest accuracy in detecting deepfake videos on both datasets.