Mamba Adaptive Anomaly Transformer with association discrepancy for time series
Authors:
Abdellah Zakaria Sellam,
Ilyes Benaissa,
Abdelmalik Taleb-Ahmed,
Luigi Patrono,
Cosimo Distante
Abstract:
Anomaly detection in time series is essential for industrial monitoring and environmental sensing, yet distinguishing anomalies from complex patterns remains challenging. Existing methods like the Anomaly Transformer and DCdetector have progressed, but they face limitations such as sensitivity to short-term contexts and inefficiency in noisy, non-stationary environments.
To overcome these issues…
▽ More
Anomaly detection in time series is essential for industrial monitoring and environmental sensing, yet distinguishing anomalies from complex patterns remains challenging. Existing methods like the Anomaly Transformer and DCdetector have progressed, but they face limitations such as sensitivity to short-term contexts and inefficiency in noisy, non-stationary environments.
To overcome these issues, we introduce MAAT, an improved architecture that enhances association discrepancy modeling and reconstruction quality. MAAT features Sparse Attention, efficiently capturing long-range dependencies by focusing on relevant time steps, thereby reducing computational redundancy. Additionally, a Mamba-Selective State Space Model is incorporated into the reconstruction module, utilizing a skip connection and Gated Attention to improve anomaly localization and detection performance.
Extensive experiments show that MAAT significantly outperforms previous methods, achieving better anomaly distinguishability and generalization across various time series applications, setting a new standard for unsupervised time series anomaly detection in real-world scenarios.
△ Less
Submitted 17 May, 2025; v1 submitted 11 February, 2025;
originally announced February 2025.
Final Architecture Specification of security, privacy, and incentive mechanisms
Authors:
Nicolai Kuntze,
Juergen Repp,
Hervais Simo Fhom,
Andreas Fuchs,
Ine-Saf Benaissa
Abstract:
In this document, we define the NADA security architecture based on refined use case scenarios, a derived high level model and security analysis. For the architecure design and verification we are applying the well known STRIDE model.
In this document, we define the NADA security architecture based on refined use case scenarios, a derived high level model and security analysis. For the architecure design and verification we are applying the well known STRIDE model.
△ Less
Submitted 17 November, 2009;
originally announced November 2009.
