references.bib

% Grid Guardian: Predictive Anomaly Detection for UK Power Grids
% BibTeX References for Aston University FYP
% Generated: January 2026
% Updated with verified author names for key papers

% ============================================================================
% GRAPH NEURAL NETWORKS FOR POWER GRIDS (KEY PAPERS)
% ============================================================================

@article{powergnn2025,
  title     = {{PowerGNN}: A Topology-Aware Graph Neural Network for Electricity Grids},
  author    = {Suri, Dhruv and Mangal, Mohak},
  journal   = {arXiv preprint arXiv:2503.22721},
  year      = {2025},
  month     = {March},
  url       = {https://arxiv.org/abs/2503.22721},
  note      = {Integrates GraphSAGE with GRUs for power system state prediction using PyTorch Geometric}
}

@inproceedings{powergraph2024,
  title     = {{PowerGraph}: A Power Grid Benchmark Dataset for Graph Neural Networks},
  author    = {Varbella, Anna and Amara, Kenza and Gjorgiev, Blazhe and El-Assady, Mennatallah and Sansavini, Giovanni},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year      = {2024},
  url       = {https://openreview.net/forum?id=qWTfCO4HvT},
  note      = {GNN-tailored benchmark for cascading failures, power flows, and optimal power flows}
}

@article{physicsgnn2023,
  title     = {Physics-Informed Graph Neural Network for Dynamic Reconfiguration of Power Systems},
  author    = {Authier, Jules and Haider, Rabab and Annaswamy, Anuradha and Dorfler, Florian},
  journal   = {arXiv preprint arXiv:2310.00728},
  year      = {2023},
  url       = {https://arxiv.org/abs/2310.00728},
  note      = {GraPhyR framework combining physics constraints with GNN for grid reconfiguration}
}

% ============================================================================
% GNN ARCHITECTURE FUNDAMENTALS (KEY PAPERS)
% ============================================================================

@inproceedings{gat2018,
  title     = {Graph Attention Networks},
  author    = {Veli{\v{c}}kovi{\'c}, Petar and Cucurull, Guillem and Casanova, Arantxa and Romero, Adriana and Li{\`o}, Pietro and Bengio, Yoshua},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2018},
  url       = {https://openreview.net/forum?id=rJXMpikCZ},
  note      = {Original GAT paper introducing attention mechanism for graphs}
}

@inproceedings{dropedge2020,
  title     = {{DropEdge}: Towards Deep Graph Convolutional Networks on Node Classification},
  author    = {Rong, Yu and Huang, Wenbing and Xu, Tingyang and Huang, Junzhou},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2020},
  url       = {https://openreview.net/forum?id=Hkx1qkrKPr},
  note      = {DropEdge alleviates overfitting and oversmoothing in deep GNNs}
}

@inproceedings{diffpool2018,
  title     = {Hierarchical Graph Representation Learning with Differentiable Pooling},
  author    = {Ying, Rex and You, Jiaxuan and Morris, Christopher and Ren, Xiang and Hamilton, William L. and Leskovec, Jure},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year      = {2018},
  url       = {https://arxiv.org/abs/1806.08804},
  note      = {DiffPool for hierarchical multi-resolution graph embeddings}
}

% ============================================================================
% ANOMALY DETECTION SURVEYS AND METHODS (KEY PAPERS)
% ============================================================================

@article{graphanomalysurvey2022,
  title     = {Graph Anomaly Detection with Graph Neural Networks: Current Status and Challenges},
  author    = {Kim, Hwan and Lee, Byung Suk and Shin, Won-Yong and Lim, Sungsu},
  journal   = {arXiv preprint arXiv:2209.14930},
  year      = {2022},
  url       = {https://arxiv.org/abs/2209.14930},
  note      = {Comprehensive survey on graph-based anomaly detection with GNNs}
}

@article{dlanomalysurvey2022,
  title     = {Deep Learning for Time Series Anomaly Detection: A Survey},
  author    = {Darban, Zahra Zamanzadeh and Webb, Geoffrey I. and Pan, Shirui and Aggarwal, Charu C. and Salehi, Mahsa},
  journal   = {ACM Computing Surveys},
  year      = {2024},
  volume    = {57},
  number    = {1},
  pages     = {1--42},
  publisher = {ACM},
  doi       = {10.1145/3691338},
  url       = {https://arxiv.org/abs/2211.05244},
  note      = {Reconstruction-based methods highly popular for unsupervised detection}
}

@inproceedings{mtadgat2020,
  title     = {{MTAD-GAT}: Multivariate Time-series Anomaly Detection via Graph Attention Network},
  author    = {Zhao, Hang and Wang, Yujing and Duan, Juanyong and Huang, Congrui and Cao, Defu and Tong, Yunhai and Xu, Bixiong and Bai, Jing and Tong, Jie and Zhang, Qi},
  booktitle = {IEEE International Conference on Data Mining (ICDM)},
  year      = {2020},
  url       = {https://arxiv.org/abs/2009.02040},
  note      = {Parallel GAT layers for feature-oriented and time-oriented attention}
}

@article{physicsautoencoder2023,
  title     = {Physics-Informed Convolutional Autoencoder for Cyber Anomaly Detection in Power Distribution Grids},
  author    = {Jabbari Zideh, Mehdi and Khushalani Solanki, Sarika},
  journal   = {arXiv preprint arXiv:2312.04758},
  year      = {2023},
  url       = {https://arxiv.org/abs/2312.04758},
  note      = {Physics constraints via Kirchhoff's law improve anomaly detection accuracy}
}

% ============================================================================
% SELF-PLAY AND AZR METHODOLOGY (KEY PAPERS)
% ============================================================================

@article{selfplaysurvey2021,
  title     = {Survey of Self-Play in Reinforcement Learning},
  author    = {DiGiovanni, Anthony and Zell, Ethan C.},
  journal   = {arXiv preprint arXiv:2107.02850},
  year      = {2021},
  url       = {https://arxiv.org/abs/2107.02850},
  note      = {Self-play effective when environment is adversarial and labeled data scarce}
}

@article{selfplaysurvey2024,
  title     = {A Survey on Self-play Methods in Reinforcement Learning},
  author    = {Zhang, Ruize and Xu, Zelai and Ma, Chengdong and Yu, Chao and Tu, Wei-Wei and Tang, Wenhao and Huang, Shiyu and Ye, Deheng and Ding, Wenbo and Yang, Yaodong and Wang, Yu},
  journal   = {arXiv preprint arXiv:2408.01072},
  year      = {2024},
  url       = {https://arxiv.org/abs/2408.01072},
  note      = {Comprehensive 2024 survey covering self-play taxonomy and applications}
}

% ============================================================================
% UNCERTAINTY QUANTIFICATION (KEY PAPERS)
% ============================================================================

@inproceedings{mcdropout2016,
  title     = {Dropout as a {Bayesian} Approximation: Representing Model Uncertainty in Deep Learning},
  author    = {Gal, Yarin and Ghahramani, Zoubin},
  booktitle = {Proceedings of The 33rd International Conference on Machine Learning (ICML)},
  pages     = {1050--1059},
  year      = {2016},
  volume    = {48},
  series    = {PMLR},
  url       = {http://proceedings.mlr.press/v48/gal16.html},
  note      = {Foundational MC Dropout paper for epistemic uncertainty quantification}
}

@article{bayesautoencoder2022,
  title     = {Bayesian Autoencoders with Uncertainty Quantification: Towards Trustworthy Anomaly Detection},
  author    = {Yong, Bang Xiang and Brintrup, Alexandra},
  journal   = {Expert Systems with Applications},
  volume    = {209},
  pages     = {118196},
  year      = {2022},
  publisher = {Elsevier},
  doi       = {10.1016/j.eswa.2022.118196},
  url       = {https://arxiv.org/abs/2202.12653},
  note      = {BAEs provide trustworthy anomaly detection via epistemic and aleatoric uncertainty}
}

% ============================================================================
% EVALUATION WITHOUT GROUND TRUTH (KEY PAPERS)
% ============================================================================

@inproceedings{unsupervisedvalidation2024,
  title     = {Towards Unsupervised Validation of Anomaly-Detection Models},
  author    = {Idan, Lihi},
  booktitle = {Proceedings of the 27th European Conference on Artificial Intelligence (ECAI)},
  year      = {2024},
  url       = {https://arxiv.org/abs/2410.14579},
  note      = {Validation methods when ground truth labels are absent}
}

% ============================================================================
% SOFTWARE AND LIBRARIES (KEY CITATIONS)
% ============================================================================

@inproceedings{pytorchgeometric2019,
  title     = {Fast Graph Representation Learning with {PyTorch Geometric}},
  author    = {Fey, Matthias and Lenssen, Jan E.},
  booktitle = {ICLR Workshop on Representation Learning on Graphs and Manifolds},
  year      = {2019},
  url       = {https://arxiv.org/abs/1903.02428},
  note      = {Primary GNN library for PyTorch with 21K+ GitHub stars}
}

@inproceedings{pygtemporal2021,
  title     = {{PyTorch Geometric Temporal}: Spatiotemporal Signal Processing with Neural Machine Learning Models},
  author    = {Rozemberczki, Benedek and Scherer, Paul and He, Yixuan and Panagopoulos, George and Riedel, Alexander and Astefanoaei, Maria and Kiss, Oliver and Beres, Ferenc and L{\'o}pez, Guzm{\'a}n and Collignon, Nicolas and Sarkar, Rik},
  booktitle = {Proceedings of the 30th ACM International Conference on Information \& Knowledge Management (CIKM)},
  pages     = {4564--4573},
  year      = {2021},
  publisher = {ACM},
  doi       = {10.1145/3459637.3482014},
  url       = {https://arxiv.org/abs/2104.07788},
  note      = {Spatiotemporal GNN library with TGCN, A3TGCN, DCRNN implementations}
}

% ============================================================================
% ADDITIONAL SUPPORTING PAPERS
% ============================================================================

@misc{circuitbreaker2024,
  title        = {Circuit Breaker Pattern},
  author       = {{Microsoft Azure}},
  year         = {2024},
  howpublished = {Azure Architecture Center},
  url          = {https://learn.microsoft.com/en-us/azure/architecture/patterns/circuit-breaker},
  note         = {Pattern for handling transient faults in distributed systems}
}

@article{rewardhacking2024,
  title   = {Reward Hacking in Reinforcement Learning},
  author  = {Weng, Lilian},
  journal = {Lil'Log},
  year    = {2024},
  url     = {https://lilianweng.github.io/posts/2024-11-28-reward-hacking/},
  note    = {Agents exploit reward functions rather than achieve intended objectives}
}

% ============================================================================
% SECONDARY REFERENCES (Anonymous - less likely to cite directly)
% ============================================================================

@article{explainablegnn2025,
  title     = {Explainable Graph Neural Networks for Power Grid Fault Detection},
  author    = {{IEEE Authors}},
  journal   = {IEEE Transactions on Power Systems},
  year      = {2025},
  doi       = {10.1109/TPWRS.2025.11088107},
  url       = {https://ieeexplore.ieee.org/document/11088107/},
  note      = {Shows GNNs exhibit remarkable precision using phasor data and topology}
}

@article{multivariatephysics2024,
  title     = {Multivariate Physics-Informed Convolutional Autoencoder for Anomaly Detection in Power Distribution Systems with High Penetration of {DERs}},
  author    = {Jabbari Zideh, Mehdi and Khushalani Solanki, Sarika},
  journal   = {arXiv preprint arXiv:2406.02927},
  year      = {2024},
  url       = {https://arxiv.org/abs/2406.02927},
  note      = {Integrates nodal power balance equations for high DER penetration}
}

@article{oversmoothing2025,
  title     = {Solving Oversmoothing in {GNNs} via Nonlocal Message Passing},
  author    = {{arXiv Authors}},
  journal   = {arXiv preprint arXiv:2512.08475},
  year      = {2025},
  month     = {December},
  url       = {https://arxiv.org/abs/2512.08475},
  note      = {Non-local message passing prevents oversmoothing up to 256 layers}
}

@article{gnntimeseriessurvey2023,
  title     = {A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection},
  author    = {{arXiv Authors}},
  journal   = {arXiv preprint arXiv:2307.03759},
  year      = {2023},
  url       = {https://arxiv.org/abs/2307.03759},
  note      = {Comprehensive survey on GNN applications to time series tasks}
}

@article{pinninverse2025,
  title     = {Ensembles of Graph and Physics-Informed Machine Learning for Scientific Modeling},
  author    = {{Springer Authors}},
  journal   = {Archives of Computational Methods in Engineering},
  year      = {2025},
  publisher = {Springer},
  doi       = {10.1007/s11831-025-10325-5},
  url       = {https://link.springer.com/article/10.1007/s11831-025-10325-5},
  note      = {Ensemble GNNs reduce MAE; stacked PINNs reduce L2 errors by 40\%+}
}

@article{hierarchicalfallback2025,
  title     = {Hierarchical Fallback Architecture for High Risk Online Machine Learning Inference},
  author    = {{arXiv Authors}},
  journal   = {arXiv preprint arXiv:2501.17834},
  year      = {2025},
  url       = {https://arxiv.org/abs/2501.17834},
  note      = {Multi-tier fallback for ML inference reliability}
}

@article{asoi2025,
  title     = {{ASOI}: Anomaly Separation and Overlap Index, an Internal Evaluation Metric for Unsupervised Anomaly Detection},
  author    = {{Springer Authors}},
  journal   = {Complex \& Intelligent Systems},
  year      = {2025},
  publisher = {Springer},
  doi       = {10.1007/s40747-025-02204-0},
  url       = {https://link.springer.com/article/10.1007/s40747-025-02204-0},
  note      = {Internal evaluation metric when ground truth is unavailable}
}

@article{distributedgnn2024,
  title     = {Distributed Graph Neural Network Training: A Survey},
  author    = {{ACM Authors}},
  journal   = {ACM Computing Surveys},
  year      = {2024},
  doi       = {10.1145/3648358},
  url       = {https://dl.acm.org/doi/10.1145/3648358},
  note      = {Survey on scaling GNN training to large graphs}
}

@software{elexondataportal,
  author = {OSUKED},
  title  = {{ElexonDataPortal}: Python Client for {BMRS} {API}},
  year   = {2023},
  url    = {https://github.com/OSUKED/ElexonDataPortal},
  note   = {Standardized parameter names, automatic date-range orchestration for UK grid data}
}