Based on the panel data of 30 provinces in China from 2013 to 2022, this paper constructs a measurement index system for the symbiotic level of digital innovation ecosystems from three dimensions: the symbiosis of digital innovation subjects, the digital innovation environment, and digital innovation interaction. This paper applies the entropy weight TOPSIS method, Dagum Gini coefficient decomposition, and spatial convergence analysis to empirically examine the symbiotic levels, regional disparities, and spatial convergence of China’s digital innovation ecosystem. The results are as follows: (i) At the national level, the symbiotic level of China’s digital innovation ecosystem has generally increased, creating a spatial distribution pattern that is “high in the east, flat in the middle, and low in the west”. (ii) From a regional perspective, the major disparities between regions are the primary factors contributing to the overall difference in the symbiotic level of China’s digital innovation ecosystem. (iii) From the perspective of σ convergence, regional disparities in the symbiotic level of the digital innovation ecosystem are constantly expanding, and uneven regional development is intensifying. (iv) From the perspective of absolute β convergence, regions with lower levels of symbiosis in the digital innovation ecosystem have a faster growth rate of symbiosis than regions with higher levels of symbiosis, and there is a certain spatial spillover effect. (v) From the perspective of conditional β convergence, economic structure and innovation application can accelerate the spatial convergence of China’s digital innovation ecosystem symbiosis to a certain extent.

As a cornerstone of smart city consumer ecosystems, accurate traffic accident risk prediction is vital for proactive safety services in intelligent transportation systems. However, current solutions often fail to integrate heterogeneous relational features (e.g., historical risk correlations) or effectively balance long-term periodic patterns with transient dynamics. To address these issues, we introduce a Cross-View Long-Short Spatial-Temporal Framework (CVLS-ST) for traffic risk prediction, which innovatively decouples long- and short-term dependencies to achieve more precise urban traffic risk prediction. Specifically, we design a cross-view attention graph network to effectively capture complex and diverse interactions across multiple relational graphs. Besides, we develop a Mamba block that incorporates state-space modeling into LSTM’s gating architecture to effectively capture global spatial semantics and temporal dependencies. Moreover, we introduce a state-inherited strategy to jointly and accurately model the long-term trends and short-term dynamics of traffic risk. Extensive experiments on two real-world datasets show that the F1-score of CVLS-ST outperforms the state-of-the-art by 20.37% on CHI and 5.83% on NYC, validating its effectiveness in complex urban environments.

The widespread usage of the Internet of Things (IoT) has raised the risks of cyber threats; thus, developing Anomaly Detection Systems (ADSs) that can adapt to evolving traffic pattern is critical. Previous studies primarily focused on offline unsupervised learning methods to safeguard ADSs, which is not applicable in practical real-world applications. In this paper, we design Adaptive NAD, an online and self-Adaptive unsupervised Network Anomaly Detection framework for security domains. A two-layer anomaly detection strategy is proposed to generate reliable high-confidence pseudo-labels. Then, an online training scheme is introduced to update Adaptive NAD by a novel threshold calculation technique. Experimental results demonstrate that Adaptive NAD achieves the lowest false alarm rate (1.33%, 0.71%, and 0.08%) and has a more than 3 times faster online inference latency compared with state-of-the-art solutions on the CIC-Darknet2020, NSL-KDD, and Edge-IIoTset datasets, respectively. The code is released at https://github.com/MyLearnCodeSpace/Adaptive-NAD.