pith. sign in

arxiv: 2605.13863 · v1 · pith:4T3EBXKMnew · submitted 2026-04-29 · 💻 cs.NE · cs.LG

Neuromorphic Graph Anomaly Detection via Adaptive STDP and Spiking Graph Neural Networks

Abdul Joseph Fofanah , Lian Wen , David Chen , Tsungcheng Yao , Kwabena Sarpong This is my paper

Pith reviewed 2026-05-15 07:15 UTC · model grok-4.3

classification 💻 cs.NE cs.LG
keywords spiking neural networksgraph anomaly detectionSTDP learningneuromorphic computingdynamic graphsLeaky Integrate-and-Fireenergy efficiency
0
0 comments X p. Extension
pith:4T3EBXKM Add to your LaTeX paper What is a Pith Number? 
\usepackage{pith}
\pithnumber{4T3EBXKM}

Prints a linked pith:4T3EBXKM badge after your title and writes the identifier into PDF metadata. Compiles on arXiv with no extra files. Learn more

The pith

A spiking neural network framework called ASTDP-GAD detects anomalies in dynamic graphs using adaptive STDP learning for neuromorphic energy efficiency.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that combining adaptive Leaky Integrate-and-Fire neurons with STDP-based plasticity in graph attention and hypergraph memory components yields accurate anomaly detection while preserving biological plausibility. A sympathetic reader would care because conventional graph anomaly methods consume too much power and struggle with temporal dynamics in networks like cybersecurity or industrial monitoring. The framework supplies theoretical bounds showing that spike encoding retains information linearly with steps and dimension, that the attention module approximates continuous functions, and that fusion steps cut score variance by up to five times.

Core claim

ASTDP-GAD unifies spiking graph neural networks with adaptive STDP learning through temporal spike encoding, LIF-based graph attention with lateral inhibition, event-driven hypergraph memory, spike-rate contrast pooling, and multi-factor anomaly fusion, delivering rigorous guarantees that encoding preserves input resolution linearly, attention approximates any continuous function, memory converges to optimal prototypes, pooling meets anomaly selection bounds, STDP converges stably, and fusion reduces variance up to 5×.

What carries the argument

The ASTDP-GAD framework that integrates adaptive LIF dynamics, STDP-inspired prototype updates, and contrast pooling to perform graph anomaly detection on spiking hardware.

Load-bearing premise

The listed theoretical approximations and convergence results continue to hold on noisy real-world dynamic graphs without major loss from discretization or hyperparameter choices.

What would settle it

A side-by-side run on a noisy dynamic graph dataset where the method's F1 score falls below standard non-spiking baselines or where measured energy per inference on neuromorphic chips exceeds conventional graph neural network implementations.

Figures

Figures reproduced from arXiv: 2605.13863 by Abdul Joseph Fofanah, David Chen, Kwabena Sarpong, Lian Wen, Tsungcheng Yao.

Figure 1
Figure 1. Figure 1: Overall architecture of the proposed ASTDP-GAD framework for neuromorphic anomaly detection. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Validation of spike encoding and LIFGAT approximation: (a) Hamming distance vs. input feature [PITH_FULL_IMAGE:figures/full_fig_p039_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Validation of EDHMM convergence (Theorem A.3), SRCGP selection, and STDP stability: (a) [PITH_FULL_IMAGE:figures/full_fig_p040_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: t-SNE visualisation of learned embeddings: (a) ground truth labels showing anomalies (red) form a [PITH_FULL_IMAGE:figures/full_fig_p040_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Anomaly score distribution showing bimodal separation: normal nodes concentrate near score 0, [PITH_FULL_IMAGE:figures/full_fig_p041_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Case study analysis: (a) true positives (n=431) show consistent high scores [PITH_FULL_IMAGE:figures/full_fig_p041_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Component contribution analysis: (a) memory scores showing multimodal prototype-based matching; [PITH_FULL_IMAGE:figures/full_fig_p042_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Training curves showing rapid convergence within 150 epochs, with validation loss closely tracking [PITH_FULL_IMAGE:figures/full_fig_p043_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Neuromorphic spike feature analysis: (a) time-to-first-spike distribution (mean 21.69); (b) total spike [PITH_FULL_IMAGE:figures/full_fig_p044_9.png] view at source ↗
read the original abstract

Anomaly detection in dynamic networks is critical for applications from cybersecurity to industrial monitoring, yet existing methods face challenges in energy efficiency, temporal precision, and adaptability. This paper introduces ASTDP-GAD, a novel Adaptive Spiking Temporal Dynamics Plasticity framework for Graph Anomaly Detection that integrates spiking graph neural networks with STDP learning for energy-efficient neuromorphic detection in dynamic networks. Our framework unifies spiking neural computation, STDP learning, and graph-based anomaly detection through the following key innovations: temporal spike graph encoding with adaptive Leaky Integrate-and-Fire (LIF) dynamics; LIF-based graph attention with lateral inhibition; event-driven hypergraph memory with STDP-inspired prototype updates; spike rate contrast pooling based on spiking irregularity; adaptive STDP layers capturing causal temporal relationships; and multi-scale temporal convolution with multi-factor anomaly fusion. Theoretical analysis provides rigorous guarantees: spike encoding preserves input information with resolution scaling linearly in simulation steps and hidden dimension; LIFGAT approximates any continuous attention function; hypergraph memory converges to optimal prototypes; contrast pooling achieves provable anomaly selection bounds; STDP learning converges stably; and multi-factor fusion produces calibrated scores with up to $5\times$ variance reduction. Extensive experiments on nine datasets on both dynamic and static graphs demonstrate superior anomaly detection accuracy while maintaining biological plausibility and energy efficiency for neuromorphic deployment.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes ASTDP-GAD, a neuromorphic framework for anomaly detection in dynamic graphs that combines adaptive spiking temporal dynamics plasticity, spiking graph neural networks, LIF-based graph attention with lateral inhibition, event-driven hypergraph memory, spike-rate contrast pooling, adaptive STDP layers, and multi-scale temporal convolution with multi-factor fusion. It asserts several theoretical guarantees including linear scaling of information preservation in spike encoding, universal approximation by LIFGAT, convergence of hypergraph memory to optimal prototypes, provable anomaly selection bounds, stable STDP convergence, and up to 5× variance reduction in calibrated scores, together with superior accuracy on nine dynamic and static graph datasets while preserving biological plausibility and energy efficiency.

Significance. If the central claims hold, the work would advance energy-efficient neuromorphic methods for graph anomaly detection by supplying both algorithmic innovations and stated theoretical bounds, potentially enabling hardware deployment with quantifiable performance advantages over conventional GNN approaches.

major comments (2)
  1. [Abstract / Theoretical Analysis] Abstract and Theoretical Analysis: the listed guarantees (linear resolution scaling in spike encoding, LIFGAT universal approximation, hypergraph memory convergence, STDP stability, and 5× variance reduction) are asserted without derivation sketches, Lipschitz constants, or discretization-error bounds; the stress-test concern that these rest on idealized continuous-time assumptions is therefore load-bearing for the central claim of rigorous guarantees.
  2. [Experiments] Experiments: the reported superiority on nine datasets is stated without dataset statistics, baseline tables, error bars, or ablation on the adaptive LIF and STDP hyperparameters; this prevents verification that the claimed gains survive the weakest assumption of transfer to noisy, irregular real-world dynamic graphs.
minor comments (2)
  1. [Notation / Methods] Define LIFGAT and all other acronyms at first use; expand the multi-factor fusion equation to show how the 5× variance reduction is obtained.
  2. [Related Work] Add a short paragraph contrasting the proposed event-driven hypergraph memory with prior spiking memory mechanisms.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help strengthen the presentation of our theoretical claims and experimental validation. We address each major point below and outline the revisions.

read point-by-point responses

  1. Referee: [Abstract / Theoretical Analysis] Abstract and Theoretical Analysis: the listed guarantees (linear resolution scaling in spike encoding, LIFGAT universal approximation, hypergraph memory convergence, STDP stability, and 5× variance reduction) are asserted without derivation sketches, Lipschitz constants, or discretization-error bounds; the stress-test concern that these rest on idealized continuous-time assumptions is therefore load-bearing for the central claim of rigorous guarantees.

    Authors: We agree that concise derivation sketches would improve accessibility. In the revised manuscript we will insert brief proof outlines for each guarantee (linear scaling of information preservation with explicit resolution bounds, LIFGAT universal approximation with Lipschitz constants, hypergraph-memory convergence to optimal prototypes, STDP stability via Lyapunov analysis, and variance reduction in the fusion step). We will also add a short appendix section supplying the requested discretization-error bounds and clarifying that all continuous-time results are stated under standard bounded-step assumptions with explicit error terms. This directly addresses the load-bearing concern without altering the core claims. revision: partial

  2. Referee: [Experiments] Experiments: the reported superiority on nine datasets is stated without dataset statistics, baseline tables, error bars, or ablation on the adaptive LIF and STDP hyperparameters; this prevents verification that the claimed gains survive the weakest assumption of transfer to noisy, irregular real-world dynamic graphs.

    Authors: We accept that the current experimental section lacks sufficient detail for independent verification. The revised version will add: (i) a table summarizing dataset statistics (nodes, edges, anomaly ratios, temporal length), (ii) full baseline tables reporting mean and standard deviation over five independent runs with error bars, and (iii) targeted ablations on the adaptive LIF time-constant and STDP learning-rate ranges. These additions will allow readers to assess robustness. Regarding explicit transfer to noisy real-world graphs, our existing benchmarks already contain irregular temporal sampling; we will expand the discussion to quantify sensitivity to added noise levels, while noting that exhaustive noisy-deployment simulations remain future work. revision: yes

Circularity Check

0 steps flagged

No significant circularity in claimed theoretical guarantees

full rationale

The paper states that theoretical analysis provides rigorous guarantees on spike encoding resolution, LIFGAT approximation of attention functions, hypergraph memory convergence to optimal prototypes, contrast pooling bounds, STDP stability, and up to 5× variance reduction from multi-factor fusion. No equations, self-citations, or derivation steps are quoted in the provided abstract that reduce any of these results to their inputs by construction, fitted parameters renamed as predictions, or load-bearing self-citations. The claims are presented as outcomes of analysis rather than tautological redefinitions or renamings, making the derivation chain self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

3 free parameters · 2 axioms · 1 invented entities

The central claims rest on standard assumptions about spiking neuron models and STDP convergence plus several adaptive mechanisms whose parameters are not shown to be fixed by prior literature.

free parameters (3)
  • adaptive LIF time constants and thresholds
    Described as adaptive but no fixed values or derivation from first principles given in abstract.
  • STDP learning rates and window parameters
    Adaptive STDP layers require rate and timing constants that are typically fitted.
  • multi-factor fusion weights
    The 5× variance reduction is stated as an outcome of fusion; weights appear chosen to achieve calibration.
axioms (2)
  • domain assumption LIF neuron dynamics and STDP update rules converge under the stated conditions
    Invoked in the theoretical analysis section of the abstract without proof sketch.
  • domain assumption Graph attention with lateral inhibition can approximate any continuous function
    Claimed for LIFGAT without reference to universal approximation theorems for spiking networks.
invented entities (1)
  • event-driven hypergraph memory no independent evidence
    purpose: Stores and updates anomaly prototypes via STDP-inspired rules
    New memory structure introduced to handle temporal graph anomalies; no independent evidence outside the framework.

pith-pipeline@v0.9.0 · 5553 in / 1573 out tokens · 39761 ms · 2026-05-15T07:15:03.212897+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

79 extracted references · 79 canonical work pages · 1 internal anchor

  1. [1]

    A generalizable anomaly detection method in dynamic graphs

    Xiao Yang, Xuejiao Zhao, and Zhiqi Shen. A generalizable anomaly detection method in dynamic graphs. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 22001–22009, 2025

    work page 2025

  2. [2]

    Graph anomaly detection in time series: A survey

    Thi Kieu Khanh Ho, Ali Karami, and Narges Armanfard. Graph anomaly detection in time series: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    work page 2025

  3. [3]

    Deep anomaly detection on attributed networks by graph update: Y

    Yifan Li, Jiayin Li, Ximeng Liu, and Li Xu. Deep anomaly detection on attributed networks by graph update: Y . li et al.Computing, 107(6):134, 2025

    work page 2025

  4. [4]

    Ge-gnn: Gated edge-augmented graph neural network for fraud detection

    Wenxin Zhang and Cuicui Luo. Ge-gnn: Gated edge-augmented graph neural network for fraud detection. IEEE Transactions on Big Data, 2025

    work page 2025

  5. [5]

    Gat-ad: Graph attention networks for contextual anomaly detection in network monitoring

    Hamid Latif-Martínez, José Suárez-Varela, Albert Cabellos-Aparicio, and Pere Barlet-Ros. Gat-ad: Graph attention networks for contextual anomaly detection in network monitoring. Computers & Industrial Engineering, 200:110830, 2025

    work page 2025

  6. [6]

    Dualbert: Fusing symbolic and temporal dynamics for high-precision log anomaly detection

    Ahmed Alzamil, Ramzi Guesmi, Amer AlGhadhban, Tawfik Guesmi, and Abdennaceur Ka- chouri. Dualbert: Fusing symbolic and temporal dynamics for high-precision log anomaly detection. IEEE Access, 2026

    work page 2026

  7. [7]

    Real-time anomaly detection in iot streams through spatiotemporal patterns

    Yazeed AlZahrani. Real-time anomaly detection in iot streams through spatiotemporal patterns. Discover Internet of Things, 6(1):4, 2026

    work page 2026

  8. [8]

    A comprehensive review of tsetlin machines: Concepts, applications, analysis, and the future

    Souraja Kundu, Shruti S Patkar, Saras Mani Mishra, Gaurav Trivedi, and Farhad Merchant. A comprehensive review of tsetlin machines: Concepts, applications, analysis, and the future. IEEE Internet of Things Journal, 2026

    work page 2026

  9. [9]

    An analytics framework for graph-based anomaly detection in healthcare time series.Healthcare Analytics, page 100447, 2026

    Emerson Yoshiaki Okano, Daniel Aloise, and Mariá CV Nascimento. An analytics framework for graph-based anomaly detection in healthcare time series.Healthcare Analytics, page 100447, 2026

    work page 2026

  10. [10]

    Mst-mfgat: An adaptive temporal convolution with feature graph attention network for time series anomaly detection

    Dan Wang, Heng Zhang, and Zhenhua Yu. Mst-mfgat: An adaptive temporal convolution with feature graph attention network for time series anomaly detection. Neurocomputing, page 132134, 2025

    work page 2025

  11. [11]

    Deanomaly: anomaly detection for multivariate time series using robust decomposition and memory-augmented diffusion models

    Hui Dou, Pengcheng Shi, Yiwen Zhang, Pengfei Chen, and Zibin Zheng. Deanomaly: anomaly detection for multivariate time series using robust decomposition and memory-augmented diffusion models. IEEE Transactions on Instrumentation and Measurement, 2025

    work page 2025

  12. [12]

    Gant-gnn: Thermodynamic-guided graph anomaly detection with synthetic feature injection for imbalanced node classification

    Abdul Joseph Fofanah, Lian Wen, David Chen, Alpha Alimamy Kamara, Tsungcheng Yao, and Albert Patrick Sankoh. Gant-gnn: Thermodynamic-guided graph anomaly detection with synthetic feature injection for imbalanced node classification. 2025. 10

    work page 2025

  13. [13]

    Neuromorphic computing: Cutting-edge advances and future

    Girish U Kamble, Chandrashekhar S Patil, Vidya V Alman, Somnath S Kundale, and Jin Hyeok. Neuromorphic computing: Cutting-edge advances and future. Recent Advances in Neuromorphic Computing, page 45, 2025

    work page 2025

  14. [14]

    Mechanical properties analysis of flexible memristors for neuromorphic computing

    Zhenqian Zhu, Jiheng Shui, Tianyu Wang, and Jialin Meng. Mechanical properties analysis of flexible memristors for neuromorphic computing. Nano-Micro Letters, 18(1):2, 2026

    work page 2026

  15. [15]

    A neuromorphic processor with on-chip learning for beyond-cmos device integration

    Hugh Greatorex, Ole Richter, Michele Mastella, Madison Cotteret, Philipp Klein, Maxime Fabre, Arianna Rubino, Willian Soares Girão, Junren Chen, Martin Ziegler, et al. A neuromorphic processor with on-chip learning for beyond-cmos device integration. Nature communications, 16(1):6424, 2025

    work page 2025

  16. [16]

    ChronoSpike: An Adaptive Spiking Graph Neural Network for Dynamic Graphs

    Md Abrar Jahin, Taufikur Rahman Fuad, Jay Pujara, and Craig Knoblock. Chronospike: An adaptive spiking graph neural network for dynamic graphs. arXiv preprint arXiv:2602.01124, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  17. [17]

    Eegsnet: A novel eeg cognitive recognition model using spiking neural network

    Jiankai Shi, Yue Zhao, Chu Wang, Hong Zeng, and Guojun Dai. Eegsnet: A novel eeg cognitive recognition model using spiking neural network. Biomedical Signal Processing and Control, 105:107610, 2025

    work page 2025

  18. [18]

    Integrated and fire spiking neuron model for improved wind speed forecasting

    Talal Alharbi, Ubaid Ahmed, Abdulelah Alharbi, and Anzar Mahmood. Integrated and fire spiking neuron model for improved wind speed forecasting. International Journal of Energy Research, 2025(1):3098062, 2025

    work page 2025

  19. [19]

    A novel label-aware global graph construction method and spiking-coded graph neural network for intelligent process fault diagnosis

    Dazi Li, Yurui Zhu, Zhihuan Song, and Hamid Reza Karimi. A novel label-aware global graph construction method and spiking-coded graph neural network for intelligent process fault diagnosis. Neurocomputing, 611:128707, 2025

    work page 2025

  20. [21]

    A cross-layer residual spiking neural network with adaptive threshold leaky integrate-and-fire neuron and learnable surrogate gradient

    Qingsong Ai, Yingnan Yang, Mincheng Cai, Kun Chen, Quan Liu, and Li Ma. A cross-layer residual spiking neural network with adaptive threshold leaky integrate-and-fire neuron and learnable surrogate gradient. Knowledge-Based Systems, 319:113575, 2025

    work page 2025

  21. [22]

    Modulated spike-time dependent plasticity (stdp)-based learning for spiking neural network (snn): A review.Neurocomputing, 618:129170, 2025

    Nazeerah Abdul Rahman and Nooraini Yusoff. Modulated spike-time dependent plasticity (stdp)-based learning for spiking neural network (snn): A review.Neurocomputing, 618:129170, 2025

    work page 2025

  22. [23]

    Deep unsupervised learning using spike-timing-dependent plasticity

    Sen Lu and Abhronil Sengupta. Deep unsupervised learning using spike-timing-dependent plasticity. Neuromorphic Computing and Engineering, 4(2):024004, 2024

    work page 2024

  23. [24]

    Anomaly detection on attributed networks via contrastive self-supervised learning

    Yixin Liu, Zhao Li, Shirui Pan, Chen Gong, Chuan Zhou, and George Karypis. Anomaly detection on attributed networks via contrastive self-supervised learning. IEEE transactions on neural networks and learning systems, 33(6):2378–2392, 2021

    work page 2021

  24. [25]

    Scaling up dynamic graph representation learning via spiking neural networks

    Jintang Li, Zhouxin Yu, Zulun Zhu, Liang Chen, Qi Yu, Zibin Zheng, Sheng Tian, Ruofan Wu, and Changhua Meng. Scaling up dynamic graph representation learning via spiking neural networks. In Proceedings of the AAAI conference on artificial intelligence, volume 37, pages 8588–8596, 2023

    work page 2023

  25. [26]

    Graph anomaly detection based on hybrid node representation learning

    Xiang Wang, Hao Dou, Dibo Dong, and Zhenyu Meng. Graph anomaly detection based on hybrid node representation learning. Neural Networks, 185:107169, 2025

    work page 2025

  26. [27]

    Anomaly detection in dynamic graphs using multiple encoding strategies via transformers

    Vishal Krishna Singh, Niharika Anand, Amrit Pal, Abishi Chowdhury, and Arjun Srivastava. Anomaly detection in dynamic graphs using multiple encoding strategies via transformers. IEEE Transactions on Consumer Electronics, 2025

    work page 2025

  27. [28]

    Rein- forcement neighborhood selection for unsupervised graph anomaly detection

    Yuanchen Bei, Sheng Zhou, Qiaoyu Tan, Hao Xu, Hao Chen, Zhao Li, and Jiajun Bu. Rein- forcement neighborhood selection for unsupervised graph anomaly detection. In 2023 IEEE International Conference on Data Mining (ICDM), pages 11–20. IEEE, 2023. 11

    work page 2023

  28. [29]

    Fiad: Graph anomaly detection framework based feature injection

    Aoge Chen, Jianshe Wu, and Hongtao Zhang. Fiad: Graph anomaly detection framework based feature injection. Expert Systems with Applications, 259:125216, 2025

    work page 2025

  29. [30]

    Label information enhanced fraud detection against low homophily in graphs

    Yuchen Wang, Jinghui Zhang, Zhengjie Huang, Weibin Li, Shikun Feng, Ziheng Ma, Yu Sun, Dianhai Yu, Fang Dong, Jiahui Jin, et al. Label information enhanced fraud detection against low homophily in graphs. In Proceedings of the ACM Web Conference 2023, pages 406–416, 2023

    work page 2023

  30. [31]

    Gad-nr: Graph anomaly detection via neighborhood reconstruction

    Amit Roy, Juan Shu, Jia Li, Carl Yang, Olivier Elshocht, Jeroen Smeets, and Pan Li. Gad-nr: Graph anomaly detection via neighborhood reconstruction. In Proceedings of the 17th ACM international conference on web search and data mining, pages 576–585, 2024

    work page 2024

  31. [32]

    Rethinking graph neural networks for anomaly detection

    Jianheng Tang, Jiajin Li, Ziqi Gao, and Jia Li. Rethinking graph neural networks for anomaly detection. In International conference on machine learning, pages 21076–21089. PMLR, 2022

    work page 2022

  32. [33]

    Alleviating structural distribution shift in graph anomaly detection

    Yuan Gao, Xiang Wang, Xiangnan He, Zhenguang Liu, Huamin Feng, and Yongdong Zhang. Alleviating structural distribution shift in graph anomaly detection. In Proceedings of the sixteenth ACM international conference on web search and data mining, pages 357–365, 2023

    work page 2023

  33. [34]

    Addressing heterophily in graph anomaly detection: A perspective of graph spectrum

    Yuan Gao, Xiang Wang, Xiangnan He, Zhenguang Liu, Huamin Feng, and Yongdong Zhang. Addressing heterophily in graph anomaly detection: A perspective of graph spectrum. In Proceedings of the ACM web conference 2023, pages 1528–1538, 2023

    work page 2023

  34. [35]

    Graph fairing convolutional networks for anomaly detection

    Mahsa Mesgaran and A Ben Hamza. Graph fairing convolutional networks for anomaly detection. Pattern Recognition, 145:109960, 2024

    work page 2024

  35. [36]

    Embedding temporal network via neighborhood formation

    Yuan Zuo, Guannan Liu, Hao Lin, Jia Guo, Xiaoqian Hu, and Junjie Wu. Embedding temporal network via neighborhood formation. page 2857–2866, New York, NY , USA, 2018. Association for Computing Machinery

    work page 2018

  36. [37]

    Yu, and Yanfang Ye

    Yuanfu Lu, Xiao Wang, Chuan Shi, Philip S. Yu, and Yanfang Ye. Temporal network embedding with micro- and macro-dynamics. page 469–478, New York, NY , USA, 2019. Association for Computing Machinery

    work page 2019

  37. [38]

    Dynamic network embedding by modeling triadic closure process

    Lekui Zhou, Yang Yang, Xiang Ren, Fei Wu, and Yueting Zhuang. Dynamic network embedding by modeling triadic closure process. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 32, 2018

    work page 2018

  38. [39]

    Transfer graph neural networks for pandemic forecasting

    George Panagopoulos, Giannis Nikolentzos, and Michalis Vazirgiannis. Transfer graph neural networks for pandemic forecasting. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 4838–4845, 2021

    work page 2021

  39. [40]

    Evolvegcn: Evolving graph convolutional networks for dynamic graphs

    Aldo Pareja, Giacomo Domeniconi, Jie Chen, Tengfei Ma, Toyotaro Suzumura, Hiroki Kaneza- shi, Tim Kaler, Tao Schardl, and Charles Leiserson. Evolvegcn: Evolving graph convolutional networks for dynamic graphs. In Proceedings of the AAAI conference on artificial intelligence, volume 34, pages 5363–5370, 2020

    work page 2020

  40. [41]

    Dy- namic spiking graph neural networks

    Nan Yin, Mengzhu Wang, Zhenghan Chen, Giulia De Masi, Huan Xiong, and Bin Gu. Dy- namic spiking graph neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 16495–16503, 2024

    work page 2024

  41. [42]

    Delay-dsgn: A dynamic spiking graph neural network with delay mechanisms for evolving graph

    Zhiqiang Wang, Jianghao Wen, and Jianqing Liang. Delay-dsgn: A dynamic spiking graph neural network with delay mechanisms for evolving graph. In Forty-second International Conference on Machine Learning, 2025

    work page 2025

  42. [43]

    Dynamic spectral graph anomaly detection

    Jianbo Zheng, Chao Yang, Tairui Zhang, Longbing Cao, Bin Jiang, Xuhui Fan, Xiao-ming Wu, and Xianxun Zhu. Dynamic spectral graph anomaly detection. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 13410–13418, 2025

    work page 2025

  43. [44]

    Slade: Detecting dynamic anomalies in edge streams without labels via self-supervised learning

    Jongha Lee, Sunwoo Kim, and Kijung Shin. Slade: Detecting dynamic anomalies in edge streams without labels via self-supervised learning. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 1506–1517, 2024. 12

    work page 2024

  44. [45]

    Anomaly detection in dynamic graphs via transformer

    Yixin Liu, Shirui Pan, Yu Guang Wang, Fei Xiong, Liang Wang, Qingfeng Chen, and Vincent CS Lee. Anomaly detection in dynamic graphs via transformer. IEEE Transactions on Knowledge and Data Engineering, 35(12):12081–12094, 2021

    work page 2021

  45. [46]

    Addressing imbalance in graph datasets: Introducing gate-gnn with graph ensemble weight attention and transfer learning for enhanced node classification

    Abdul Joseph Fofanah, David Chen, Lian Wen, and Shaoyang Zhang. Addressing imbalance in graph datasets: Introducing gate-gnn with graph ensemble weight attention and transfer learning for enhanced node classification. Expert Systems with Applications, 255:124602, 2024

    work page 2024

  46. [47]

    Eatsa-gnn: Edge-aware and two-stage attention for enhancing graph neural networks based on teacher–student mechanisms for graph node classification

    Abdul Joseph Fofanah and Alpha Omar Leigh. Eatsa-gnn: Edge-aware and two-stage attention for enhancing graph neural networks based on teacher–student mechanisms for graph node classification. Neurocomputing, 612:128686, 2025

    work page 2025

  47. [48]

    Graph evidential learning for anomaly detection

    Chunyu Wei, Wenji Hu, Xingjia Hao, Yunhai Wang, Yueguo Chen, Bing Bai, and Fei Wang. Graph evidential learning for anomaly detection. In Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V .2, pages 3122–3133, 2025

    work page 2025

  48. [49]

    Context-aware inductive knowledge graph completion with latent type constraints and subgraph reasoning

    Muzhi Li, Cehao Yang, Chengjin Xu, Zixing Song, Xuhui Jiang, Jian Guo, Ho-fung Leung, and Irwin King. Context-aware inductive knowledge graph completion with latent type constraints and subgraph reasoning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 12102–12111, 2025

    work page 2025

  49. [50]

    Evolution of malicious social bot detection: From individual profiling to group analysis and beyond

    Yina Liu, Shuai Xu, Yicong Li, and Shuo Yu. Evolution of malicious social bot detection: From individual profiling to group analysis and beyond. Journal of Social Computing, 6(3):258–284, 2025

    work page 2025

  50. [51]

    A-nids: adaptive network intrusion detection system based on clustering and stacked ctgan

    Chao Zha, Zhiyu Wang, Yifei Fan, Bing Bai, Yinjie Zhang, Sainan Shi, and Ruyun Zhang. A-nids: adaptive network intrusion detection system based on clustering and stacked ctgan. IEEE Transactions on Information Forensics and Security, 2025

    work page 2025

  51. [52]

    A comprehensive survey of dynamic graph neural networks: Models, frameworks, benchmarks, experiments and challenges

    ZhengZhao Feng, Rui Wang, TianXing Wang, Mingli Song, Sai Wu, and Shuibing He. A comprehensive survey of dynamic graph neural networks: Models, frameworks, benchmarks, experiments and challenges. IEEE Transactions on Knowledge and Data Engineering, 2025

    work page 2025

  52. [53]

    Dynamic graph representation learning with neural networks: A survey

    Leshanshui Yang, Clément Chatelain, and Sébastien Adam. Dynamic graph representation learning with neural networks: A survey. Ieee Access, 12:43460–43484, 2024

    work page 2024

  53. [54]

    From droplet to diagnosis: Spatio-temporal pattern recognition in drying biofluids

    Anusuya Pal, Amalesh Gope, and Miho Yanagisawa. From droplet to diagnosis: Spatio-temporal pattern recognition in drying biofluids. Advanced Intelligent Systems, 8(2):e202500550, 2026

    work page 2026

  54. [55]

    Mage: Multiperiodic adaptive graph evolu- tion guided anomaly detection in industrial iot

    Weixu Wang, Xiaobo Zhou, Tie Qiu, and Lei Wang. Mage: Multiperiodic adaptive graph evolu- tion guided anomaly detection in industrial iot. IEEE Transactions on Industrial Informatics, 2025

    work page 2025

  55. [56]

    Graph neural networks: architectures, applications, and future directions

    Valerio Ponzi and Christian Napoli. Graph neural networks: architectures, applications, and future directions. IEEE Access, 2025

    work page 2025

  56. [57]

    Trustaware-gnn: Graph neural network-based trust management for iot anomaly detection

    Kamran Ahmad Awan, Ikram Ud Din, Ahmad Almogren, Zhu Han, and Mohsen Guizani. Trustaware-gnn: Graph neural network-based trust management for iot anomaly detection. IEEE Internet of Things Journal, 2025

    work page 2025

  57. [58]

    Spike-timing-dependent plasticity and synaptic consolidation in hfo 2 memristors for adaptive neuromorphic computing

    Mostafa Shooshtari, Saeideh Pahlavan, Teresa Serrano Gotarredona, and Bernabé Linares- Barranco. Spike-timing-dependent plasticity and synaptic consolidation in hfo 2 memristors for adaptive neuromorphic computing. Neuromorphic Computing and Engineering, 5(4):044008, 2025

    work page 2025

  58. [59]

    Repetitive magnetic stimulation induces plasticity of excitatory synapses through cooperative pre-and postsynaptic activity

    Christos Galanis, Nicholas Hananeia, Maximilian Lenz, Mohammadreza Vasheghani Farahani, Peter Jedlicka, and Andreas Vlachos. Repetitive magnetic stimulation induces plasticity of excitatory synapses through cooperative pre-and postsynaptic activity. Brain Stimulation, 2025

    work page 2025

  59. [60]

    Unsupervised local learning based on voltage-dependent synaptic plasticity for resistive and ferroelectric synapses

    Nikhil Garg, Ismael Balafrej, Joao Henrique Quintino Palhares, Laura Bégon-Lours, Davide Florini, Donato Francesco Falcone, Tommaso Stecconi, Valeria Bragaglia, Bert Jan Offrein, Jean-Michel Portal, et al. Unsupervised local learning based on voltage-dependent synaptic plasticity for resistive and ferroelectric synapses. Communications Materials, 2026. 13

    work page 2026

  60. [61]

    Emerging neuromorphic devices-compatible snn hardware with adaptive stdp and validation using novel cmos neuron- synapse circuits

    Min Jee Kim, Hyung-Min Lee, Yeonjoo Jeong, and Joon Young Kwak. Emerging neuromorphic devices-compatible snn hardware with adaptive stdp and validation using novel cmos neuron- synapse circuits. IEEE Access, 2025

    work page 2025

  61. [62]

    Spiking neural networks: a comprehensive survey of training methodologies, hardware implementations and applications

    Ameer Hamza Khan, Xinwei Cao, Chunbo Luo, Shiqing Zhang, Wenping Guo, Vasilios N Kat- sikis, and Shuai Li. Spiking neural networks: a comprehensive survey of training methodologies, hardware implementations and applications. Artificial Intelligence Science and Engineering, 1(3):175–207, 2025

    work page 2025

  62. [63]

    Digital design of a spatial-pow-stdp learning block with high accuracy utilizing pow cordic for large-scale image classifier spatiotemporal snn

    Mohammad Kazem Bahrami and Soheila Nazari. Digital design of a spatial-pow-stdp learning block with high accuracy utilizing pow cordic for large-scale image classifier spatiotemporal snn. Scientific reports, 14(1):3388, 2024

    work page 2024

  63. [64]

    Fault diagnosis using liquid state machine with spiking- timing-dependent plasticity learning rule

    Yi Wan, Shaoping Wang, and Di Liu. Fault diagnosis using liquid state machine with spiking- timing-dependent plasticity learning rule. Expert Systems with Applications, 271:126736, 2025

    work page 2025

  64. [65]

    Spike-timing dependent plasticity learning of small spiking neural network for image recognition

    A V Kurbako, DM Ezhov, VI Ponomarenko, and MD Prokhorov. Spike-timing dependent plasticity learning of small spiking neural network for image recognition. The European Physical Journal Special Topics, 234(15):3659–3668, 2025

    work page 2025

  65. [66]

    Neuronal competition groups with supervised stdp for spike-based classification

    Gaspard Goupy, Pierre Tirilly, and Ioan Marius Bilasco. Neuronal competition groups with supervised stdp for spike-based classification. Advances in Neural Information Processing Systems, 37:106295–106314, 2024

    work page 2024

  66. [67]

    Stsf: Spiking time sparse feedback learning for spiking neural networks

    Ping He, Rong Xiao, Chenwei Tang, Shudong Huang, Jiancheng Lv, and Huajin Tang. Stsf: Spiking time sparse feedback learning for spiking neural networks. IEEE Transactions on Neural Networks and Learning Systems, 36(6):11479–11492, 2025

    work page 2025

  67. [68]

    Spiking graph neural network on riemannian manifolds

    Li Sun, Zhenhao Huang, Qiqi Wan, Hao Peng, and Philip S Yu. Spiking graph neural network on riemannian manifolds. Advances in Neural Information Processing Systems, 37:34025–34055, 2024

    work page 2024

  68. [69]

    Dynamic reactive spiking graph neural network

    Han Zhao, Xu Yang, Cheng Deng, and Junchi Yan. Dynamic reactive spiking graph neural network. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 16970–16978, 2024

    work page 2024

  69. [70]

    Dynamic graph neural network-based fraud detectors against collaborative fraudsters

    Lingfei Ren, Ruimin Hu, Dengshi Li, Yang Liu, Junhang Wu, Yilong Zang, and Wenyi Hu. Dynamic graph neural network-based fraud detectors against collaborative fraudsters. Knowledge-Based Systems, 278:110888, 2023

    work page 2023

  70. [71]

    Ahgt-dfd: Adaptive hierarchical graph transformer for dynamic fraud detection in blockchain networks

    Jianbin Gao, Befoum Stephane Richard, Hu Xia, Kombou Victor, Eyezo’o Benjamin Fabien, and Qi Xia. Ahgt-dfd: Adaptive hierarchical graph transformer for dynamic fraud detection in blockchain networks. IEEE Transactions on Dependable and Secure Computing, 2025

    work page 2025

  71. [72]

    Nowhere to h 2 ide: fraud detection from multi-relation graphs via disentangled homophily and heterophily identification

    Chao Fu, Guannan Liu, Kun Yuan, and Junjie Wu. Nowhere to h 2 ide: fraud detection from multi-relation graphs via disentangled homophily and heterophily identification. IEEE Transactions on Knowledge and Data Engineering, 37(3):1380–1393, 2024

    work page 2024

  72. [73]

    Cost-sensitive gnn-based imbalanced learning for mobile social network fraud detection

    Xinxin Hu, Haotian Chen, Hongchang Chen, Shuxin Liu, Xing Li, Shibo Zhang, Yahui Wang, and Xiangyang Xue. Cost-sensitive gnn-based imbalanced learning for mobile social network fraud detection. IEEE Transactions on Computational Social Systems, 11(2):2675–2690, 2023

    work page 2023

  73. [74]

    Multi-relational graph representation learning for financial statement fraud detection

    Chenxu Wang, Mengqin Wang, Xiaoguang Wang, Luyue Zhang, and Yi Long. Multi-relational graph representation learning for financial statement fraud detection. Big Data Mining and Analytics, 7(3):920–941, 2024

    work page 2024

  74. [75]

    Improving financial statement fraud detection: A large language model processing approach

    Yue Yu, Zhen Wu, Yanni Han, Ying Ding, and Wenqi Wei. Improving financial statement fraud detection: A large language model processing approach. ACM Transactions on Internet Technology, 2026

    work page 2026

  75. [76]

    Theoretical neuroscience: computational and mathematical modeling of neural systems

    Peter Dayan and Laurence F Abbott. Theoretical neuroscience: computational and mathematical modeling of neural systems. MIT press, 2005. 14

    work page 2005

  76. [77]

    Networks of spiking neurons: the third generation of neural network models

    Wolfgang Maass. Networks of spiking neurons: the third generation of neural network models. Neural networks, 10(9):1659–1671, 1997

    work page 1997

  77. [78]

    Stochastic approximation and recursive algorithms and applications

    Harold J Kushner and G George Yin. Stochastic approximation and recursive algorithms and applications. Springer, 2003

    work page 2003

  78. [79]

    Stochastic approximation methods for constrained and unconstrained systems

    Harold Joseph Kushner and Dean S Clark. Stochastic approximation methods for constrained and unconstrained systems. Springer Science & Business Media, 2012. 15 Contents 1 Introduction 2 2 Preliminaries 3 3 The Proposed Method 4 3.1 Temporal Spike Graph Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3.2 Leaky Integrate-and-Fire Graph Atte...

    work page 2012

  79. [80]

    Delay-DSGN [42] incorporates delay mechanisms into spiking GNNs for evolving graphs

    introduced an adaptive spiking GNN for dynamic graphs with learnable temporal dynamics. Delay-DSGN [42] incorporates delay mechanisms into spiking GNNs for evolving graphs. Dynamic Spiking Graph Neural Networks [41] and Dynamic Reactive Spiking Graph Neural Networks [69] explore adaptive spiking mechanisms for temporal graph processing. Despite these adva...

    work page arXiv 2000