RADAR

Welcome to the RADAR documentation!

Robust Anomaly Detection And Recognition (RADAR) is a unified platform for anomaly detection that integrates diverse approaches and libraries from the literature, alongside innovative model variants. RADAR aims to provide a flexible and extensible framework covering methods from classical statistical techniques to advanced Transformer-based architectures, including support for Federated Learning in distributed environments.

Features

  • Integration of classical and state-of-the-art anomaly detection methods.

  • Transformer-based models for time series and high-dimensional data.

  • Support for Federated Learning, enabling privacy-preserving distributed training.

  • Extensible and modular design for adding custom models and evaluation strategies.

Supported Methods

Specifically, RADAR includes:

  • Classical methods for static data: integration with PyOD and Scikit-learn.

  • Time series and deep learning models: integration with libraries such as TSFE-DL.

  • Representative Transformer models: Informer, Autoformer, and Vanilla Transformer (implemented within the time_series/ folder).

  • Federated anomaly detection: integration with flex-anomalies, developed as part of the Flexible platform.

Library / Model

Brief Description

Citation

PyOD

Collection of classical algorithms for anomaly detection on static data.

PyOD

Scikit-learn

Traditional machine learning methods applied to anomaly detection.

Scikit-learn

TSFE-DL

Framework for anomaly detection in time series using deep learning.

TSFE-DL

Informer

Transformer-based model optimized for long time series forecasting and anomaly detection.

Informer

Autoformer

Transformer specialized in time series forecasting and pattern detection.

Autoformer

Vanilla Transformer

Base Transformer implementation applied to anomaly detection.

Attention Is All You Need

flex-anomalies

Library for anomaly detection in Federated Learning environments, part of the Flexible platform.

flex-anomalies

Getting Started

To get started with RADAR, check the modules section below or explore the tutorials for step-by-step guidance.

Contents: