STEAMCODER: Spatial and Temporal Adaptive Dynamic Convolution Autoencoder for Anomaly Detection

Peng Xu, Honglu Gan, Hua Fu, Zhongbao Zhang

Abstract: Highlights•Design a Spatial Feature Extraction Algorithm based on K-Means.•Design a Temporal Feature Extraction Algorithm based on LSTM.•Convert the data into a spatial–temporal anomaly feature matrix.•Use a dynamic convolution autoencoder to analyze the matrix and detect anomalies.•Verify the superiority of our method in adaptability and accuracy to SOTA methods.