This paper presents DS-CTAN, a lightweight two-stream network for facial expression recognition in edge-based distributed systems. Designed for heterogeneous devices such as mobile phones, Jetson, and Raspberry Pi with TPU, it combines spatial feature extraction with temporal attention to capture dynamic facial movements. Using transfer learning and knowledge distillation, the model achieves high accuracy with low computational cost. Evaluated on multiple public datasets and a self-built cross-domain benchmark, DS-CTAN enables real-time, on-device inference with fast response and low power consumption. By performing inference locally, it reduces latency and data transmission, making it suitable for distributed AI applications like smart environments, human-computer interaction, and edge-driven affective computing. The framework provides a practical solution for deploying efficient vision models in resource-constrained, distributed architectures.
In recent years, the integration of science and art has continuously promoted the dynamic inheritance of traditional culture into a new stage. This article focuses on the mechanism optimization and adaptation decisions of the active inheritance of Hetu Luoshu under the background of AI empowerment, attempting to break the one-way link of display replication and explore the multi-layer feedback between symbol semantics, user behavior, and AI self-learning. The team relies on multi-source data collection and simulation experiments to conduct in-depth modeling of key variables such as symbol features and interaction strength, analyze the adaptation peaks, valleys, mismatch faults of AI decision models under different mechanisms, and identify real bottlenecks and innovative points in the process of active inheritance. The study simultaneously introduces dynamic intervention paths such as user stratification and expert feedback and constructs an adaptive decision-making system that is both closed-loop and behavior driven.
This study uses Sentinel satellite data to estimate forest quality over a large area, focusing on Beijing. By combining ground survey data with remote sensing, a random forest model predicts forest parameters. The results show a correlation coefficient of 0.60-0.76 and a relative root mean square error of 0.09-0.39. Average tree height and diameter at breast height (DBH) had the highest accuracy (75%-80%), followed by canopy density and plant number density (68%-75%). The spatial agreement between predicted and actual forest quality indicates the model's effectiveness.
This study uses Sentinel satellite data to estimate forest quality over a large area, focusing on Beijing. By combining ground survey data with remote sensing, a random forest model predicts forest parameters. The results show a correlation coefficient of 0.60-0.76 and a relative root mean square error of 0.09-0.39. Average tree height and diameter at breast height (DBH) had the highest accuracy (75%-80%), followed by canopy density and plant number density (68%-75%). The spatial agreement between predicted and actual forest quality indicates the model's effectiveness.
With the comprehensive and in-depth implementation of the “going global” strategy of Chinese culture, research on the overseas dissemination of Chinese literature has gradually become a prominent discipline. This article first defines the essence of overseas translation of Chinese literature, and based on Brado's 7W communication model, combined with the current situation of overseas dissemination of Chinese literature, analyzes the motivation of translation ecology research on overseas dissemination of Chinese literature. Through simulation experiments and actual data analysis, the differences in the accuracy of individual and joint regularization methods in small sample high-dimensional feature selection of deep learning models were compared. The research on the overseas dissemination of Chinese literature is still in its infancy, and conducting research and analysis on it is of great significance.
