Code underlying the publication: "RESTAD: Reconstruction and Similarity Transformer for time series Anomaly Detection"

This repository contains the official implementation of RESTAD (REconstruction and Similarity-based Transformer for time series Anomaly Detection), a novel framework that integrates reconstruction error with Radial Basis Function (RBF) similarity scores to enhance sensitivity to subtle anomalies. RESTAD leverages a Transformer architecture with an embedded RBF layer to synergistically detect anomalies in time series data, outperforming existing baselines on multiple benchmark datasets.