Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{208430,
        author = {Krishna Dhara Syava and Nikhil Ranjan},
        title = {Predictive Cloud Anomaly Detection Using a Hybrid Transformer-LSTM Neural Network Architecture},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {4},
        pages = {2054-2058},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=208430},
        abstract = {Multi-tenant cloud computing infrastructures remain vulnerable to multi-stage advanced persistent threats (APTs) where standard signature-based detection systems yield high error configurations. This paper evaluates a hybrid deep learning model that integrates sequential Long Short-Term Memory (LSTM) layers with a Transformer multi-head self-attention mechanism for binary and multi-class network anomaly classification. The proposed network ingests time-series telemetry data from Virtual Private Cloud (VPC) flow logs and stateless API gateway infrastructure. Evaluated against the CICIDS2017 and UNSW-NB15 public benchmark datasets, the architecture demonstrates an empirical accuracy of 99.42%, a precision of 98.91%, and an F1-score of 99.16%. The recorded inference pipeline latency averages 14.2 milliseconds, providing computational throughput sufficient to initiate automated remediation patterns prior to network payload compromise.},
        keywords = {Cloud security, intrusion detection, long short-term memory, predictive analytics, self-attention mechanisms.},
        month = {September},
        }