Intelligent deep fusion network for urban traffic flow anomaly identification

This paper presents a novel deep learning architecture for identifying outliers in the context of intelligent transportation systems. The use convolutional neural network with an efficient decomposition strategy is explored to find anomalous behavior urban traffic flow data. data set decomposed into similar clusters, each containing homogeneous used cluster. In this way, different models are trained, learned from highly correlated A merging finally fuse results obtained models. To validate performance proposed framework, intensive experiments were conducted on show that our system outperforms competition several accuracy criteria.