Pith. sign in

REVIEW 3 major objections 4 minor 56 references

A topology-aware graph reinforcement learning framework, combining structure-aware state encoding with policy-adaptive graph updates, reports better throughput, latency, and link balance than four existing graph-routing baselines on the GEA

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-05 05:43 UTC pith:TCGYOYDJ

load-bearing objection The PAGU graph-rewriting module makes Table 1 comparisons not apples-to-apples; otherwise the paper is a standard GRL routing method with under-specified details. the 3 major comments →

arxiv 2509.04973 v1 pith:TCGYOYDJ submitted 2025-09-05 cs.LG

Topology-Aware Graph Reinforcement Learning for Dynamic Routing in Cloud Networks

Yuxi Wang , Heyao Liu , Guanzi Yao , Nyutian Long , Yue Kang This is my paper
classification cs.LG
keywords graph reinforcement learningtopology awarenessdynamic routingcloud networksstructure-aware state encodingpolicy-adaptive graph updateGEANTtraffic engineering
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper argues that routing decisions in cloud networks improve when the reinforcement learning agent is explicitly aware of the network's evolving topology and is allowed to adjust that topology as part of its policy. To test this, it builds a graph-RL framework with two additions: a Structure-Aware State Encoding module (SASE) that mixes graph convolutions with positional embeddings and attention-weighted neighbor aggregation, and a Policy-Adaptive Graph Update module (PAGU) that prunes and adds edges based on policy shift and reward. On the GEANT research-network topology, the combined model reports higher average throughput (9.81 vs 9.54 for the best baseline), lower average latency (27.3 vs 29.7), lower maximum link utilization (74.2% vs 76.9%), and higher cumulative reward (288.9 vs 275.1). The paper's central claim is that these two mechanisms complement each other—SASE gives the agent structural awareness, PAGU gives it structural adaptability—and that together they beat current graph-RL routing baselines on a realistic dynamic network.

Core claim

The paper's central claim is that the two proposed mechanisms are complementary and jointly responsible for the gains. In the comparative experiment, the full model reports throughput 9.81 against 9.54 for the best graph-RL baseline, average latency 27.3 against 29.7, maximum link utilization 74.2% against 76.9%, and reward 288.9 against 275.1. The ablation shows that SASE alone lifts the plain baseline (throughput 9.02→9.34, latency 33.5→30.1, reward 253.7→267.6), PAGU alone also helps but less on latency (31.4, reward 262.3), and the two together give the best numbers (9.81, 27.3, 74.2%, 288.9). The paper therefore argues that structure-aware state representation and policy-adaptive graph

What carries the argument

SASE (Structure-Aware State Encoding) is the state-representation module: it runs multi-layer graph convolution over the adjacency matrix, augments each node's representation with a positional embedding based on shortest-path distances, and then aggregates neighbor features with attention weights. PAGU (Policy-Adaptive Graph Update) is the structural-adaptation module: it measures how much the policy's action distribution changed between time steps, scores each edge's importance by reward-weighted scheduling frequency, keeps edges above a retention threshold (0.6), introduces candidate edges above an introduction threshold (0.4), and returns a new adjacency matrix to the encoder and environm

Load-bearing premise

The load-bearing premise is that modifying the network's adjacency structure—removing and adding links according to policy feedback—is a legitimate routing action in a dynamic cloud network; if real routers cannot edit their physical topology, the measured gains may reflect the agent choosing a favorable graph rather than routing better on the given graph.

What would settle it

Run the full TAGRL model on the GEANT topology with the adjacency matrix frozen (PAGU disabled for edge additions/removals but SASE and policy training identical), and compare against the best baseline; if the model does not preserve the reported margin (or ties with SASE-only), the PAGU gains are an artifact of topology editing rather than an improvement in routing.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Routing agents that combine structural encoding with adaptive graph updates can cut maximum link utilization by roughly three percentage points over the best baseline, leaving more headroom before congestion.
  • Ablation results imply that the modules can be adopted incrementally: adding SASE for state representation or PAGU for topological adjustment each yields measurable gains, and the combined upgrade is larger than either alone.
  • Because the framework is graph-generic, the same pair of modules could transfer to other dynamic resource-allocation tasks where an agent controls a network, such as 5G edge computing, IoT networks, and data-center energy management.
  • Sensitivity experiments suggest practical deployment guidelines: a discount factor around 0.96, graph density above a 0.6 retention ratio, and 128-dimensional node features balance performance and stability.
  • If PAGU is taken literally as topology editing, the method also implies that routing policy and network topology can be co-optimized rather than treating the graph as fixed.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A natural next experiment is to give every baseline the same graph-edit action space; if baselines close the gap once they can also rewire the network, the reported advantage comes from the extra action space, not from the SASE encoder.
  • The 90–110% edge-count bound in PAGU can be reinterpreted as a 'topology flexibility budget'; exposing that budget as a tunable parameter and measuring performance versus rewiring cost would reveal how much of the gain depends on the freedom to add edges.
  • The GEANT graph is a single, moderately sized WAN topology; the framework's robustness claim would be stronger with a transfer test across other network architectures (data-center fat-trees, sparser research networks) to see whether the SASE/PAGU advantage persists.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes TAGRL, a graph reinforcement learning framework for dynamic routing in cloud networks. It combines a Structure-Aware State Encoding (SASE) module that fuses multi-layer graph convolution with positional embeddings and attention-weighted neighbor aggregation, and a Policy-Adaptive Graph Update (PAGU) module that rewrites the adjacency matrix based on policy deviation and reward feedback. The method is evaluated on the GEANT topology against four baselines (GDDR, DeepCQ+, CFR-RL, GRL-TE) and via ablation and sensitivity experiments. The authors report improvements in throughput, latency, maximum link utilization, and cumulative reward, concluding that the proposed method outperforms existing graph-RL routing models.

Significance. If the claims were supported, the work would be a useful step toward topology-aware RL routing, particularly the idea of using structural encoding to improve state representation under dynamic network conditions. The paper, however, provides no code, no reproducibility artifacts, and no statistical evidence. The central comparative claim rests on single numbers in Tables 1 and 2, and the experimental design allows the agent to alter the graph structure, so the reported gains may not reflect routing performance on the original network. The core equations also omit definitions of the very quantities they depend on. The work is therefore best viewed as an early idea requiring substantial validation before its significance can be judged.

major comments (3)
  1. [§III-B, Eqs. (8)–(9); §IV-B] The PAGU module rewrites the environment's adjacency matrix. Eq. (8) drops edges whose importance w_ij is below retention threshold tau=0.6 and adds candidate edges whose policy relevance phi_ij is above introduction threshold gamma=0.4; Eq. (9) writes A_{t+1} back into the structure encoder and environment module. With edge count constrained to 90%–110% of the initial graph (§IV-B), the agent is not solving the same routing problem as the baselines. The Table 1 gains could therefore be due to selecting a more favorable graph rather than making better routing decisions on the original GEANT topology. Real routers cannot physically add or remove links at routing timescale. Please either restrict PAGU to a interpretive/auxiliary role and evaluate routing on the fixed topology, give all baselines access to the same graph-edit actions, or explicitly reframe the problem as joint topology desi
  2. [§III-A Eq. (5); §III-B Eqs. (7)–(8)] Core quantities in the method are never defined. Eq. (5) uses alpha_ij as an 'attention-based weight coefficient' but no attention computation is given. Eq. (7) depends on f_ij^{(t)}, described only as 'scheduling frequency', and Eq. (8) depends on phi_ij^{(t)}, the 'policy relevance score', which is never defined. Eq. (6) also uses notation loosely (state tc instead of s_t). Without explicit definitions and a concrete algorithm, SASE and PAGU are not reproducible and the comparative claim cannot be independently verified. Please provide complete equations, normalization steps, and pseudocode for the attention weights and the PAGU scoring mechanism.
  3. [§IV-B, §IV-C, Tables 1–2, Fig. 4] The empirical evidence is statistically weak. Tables 1 and 2 report single values with no standard deviations, no number of independent seeds, no repetitions, and no significance tests. The hyperparameter description in §IV-B sets discount factor gamma=0.95, while §IV-C3 reports the optimal gamma as 0.96, suggesting that evaluation was used for model selection; no train/validation/test split is described. The claim that the proposed method 'outperforms existing graph reinforcement learning models' needs multiple seeds, error bars, statistical tests, and an explicit held-out protocol. This is a load-bearing issue because the entire paper is built on empirical superiority.
minor comments (4)
  1. [§IV-C4] The fourth subsection heading is duplicated as 'The impact of discount factor changes on strategy stability'; it should refer to graph sparsity/retention ratio.
  2. [§III-A, Eq. (6)] Eq. (6) has a typo: 'the action probability distribution of the state tc' should be 'state s_t'. Also clarify the relationship between policy parameter theta_t and the action distribution.
  3. [Figures 4–6] The figures are referenced but no axes, units, or error bars are described in the text. Please ensure the final figures are legible and include confidence intervals or variance bands.
  4. [Introduction, References] The introduction cites numerous works on LLMs, finance, and medical networks (Refs. [8]–[21]) that are not directly connected to graph-RL routing. These do not strengthen the motivation and should be replaced with networking-RL literature or removed.

Circularity Check

1 steps flagged

Evaluation loop: PAGU rewrites the network graph before scoring, so Table 1 compares different routing problems.

specific steps
  1. self definitional [Section III-B, Eqs. (7)-(9); Section IV-C-1, Table 1]
    "Finally, the PAGU module returns the updated adjacency matrix A_{t+1} to the structure encoder and environment module to complete the periodic reconstruction of the graph state: ... A_{t+1}(i,j)=1 if (i,j)∈E_{t+1}, else 0. ... the number of edges in the graph is kept within 90% to 110% of the initial graph."

    The agent is allowed to edit the adjacency matrix that defines the routing environment (Eq. 9). The edge-retention score w_ij in Eq. (7) is proportional to the global reward r_t, so edges contributing to high reward are kept and low-reward or congested edges can be dropped. The performance metrics in Table 1 (throughput, latency, max link utilization, reward) are then measured on this agent-modified topology, not on the fixed GEANT graph used by the baselines. Therefore the reported improvements, e.g. 'reducing maximum link utilization by nearly 10 percentage points through PAGU', are partly achieved by deleting congested links—a topology edit that baselines cannot perform. The comparison is between different routing problems, so the central claim reduces, in part, to a self-selected envir

full rationale

The paper is primarily an empirical comparison rather than a formal derivation, so classical circularity of the 'prediction equals fitted input' type is limited. However, the PAGU module creates a self-referential evaluation loop: the agent writes its chosen adjacency matrix back into the environment module (Eq. 9), and then the reported throughput, latency, link utilization, and reward are computed on that modified graph. Since the baselines do not have PAGU, they are evaluated on the original GEANT graph while the proposed agent is evaluated on a graph it has edited, up to a 10% edge-count change. This makes the headline superiority in Table 1 partly an artifact of changing the problem itself. The thresholds (0.6, 0.4, 0.1) are stated without a held-out split, and the discount-factor sensitivity analysis reports an optimum of 0.96 while the main setup uses 0.95, adding reproducibility concerns. Nonetheless, SASE is an independent structure-aware encoding module and the ablation shows gains even without PAGU, so the central claim is not entirely forced by construction. The circularity score of 4 reflects a significant but partial evaluation circularity.

Axiom & Free-Parameter Ledger

11 free parameters · 6 axioms · 0 invented entities

The central empirical claim rests on hand-set thresholds (0.6, 0.4, 0.1, 90%-110%), on two unexplained quantities (alpha_ij and phi_ij), and on the premise that modifying the graph structure during evaluation is a legitimate routing action. No trained parameters, seeds, or code are released, so the reported gains cannot be separated from these tuning choices.

free parameters (11)
  • edge_retention_threshold_tau = 0.6
    Set in Section IV-B. Controls which edges are kept in the PAGU update; the reported gains depend on this hand-chosen value.
  • edge_introduction_threshold_gamma = 0.4
    Set in Section IV-B. Controls which candidate edges are added; the candidate-edge relevance phi_ij is never defined.
  • action_deviation_threshold = 0.1
    Set in Section IV-B. Triggers PAGU updates based on policy behavior shifts.
  • edge_count_bounds = 90%-110% of initial graph
    Set in Section IV-B. Keeps graph size near the initial value during rewiring.
  • discount_factor_gamma_RL = 0.95 (0.96 claimed optimal in Figure 4)
    Tuned on the evaluation setup without a reported held-out split.
  • attention_weights_alpha_ij = not reported (learned)
    Eq (5) depends on these coefficients, but no attention mechanism equation or training detail is given.
  • policy_relevance_score_phi_ij = not defined
    Eq (8) uses phi_ij to add candidate edges; the paper never specifies what it is or how it is computed.
  • hidden_dimension = 128
    Architecture choice in Section IV-B.
  • node_feature_dimension = 64
    Architecture choice in Section IV-B.
  • positional_embedding_dimension = 16
    Architecture choice in Section IV-B.
  • entropy_regularization_coefficient = 0.01
    Hyperparameter in Section IV-B affecting exploration and stability.
axioms (6)
  • standard math The standard GCN update in Eq (2) is an appropriate encoder for routing state.
    Eq (2) is the standard GCN propagation rule, but the paper does not justify its suitability for dynamic routing or cite its origin.
  • domain assumption The network can be represented by a fixed node set and an adjacency matrix that the agent may modify.
    Section III-B Eqs (8)-(9) and Section IV-B edge-count bounds assume physical links can be dropped and added by the policy.
  • domain assumption Shortest-path-based positional embeddings capture structural roles relevant to routing.
    Eq (3) in Section III-A introduces positional embeddings without evidence that shortest-path distance is the right structural signal.
  • ad hoc to paper A scalar global reward r_t at each timestep is sufficient to assign per-edge importance w_ij.
    Eq (7) defines w_ij using r_t times edge scheduling frequency; no justification is given for this specific importance model.
  • domain assumption GEANT with synthetic traffic approximates dynamic cloud server environments.
    Section IV-A states this suitability as a premise for generalizing the empirical results.
  • ad hoc to paper The policy deviation Delta_pi_t in Eq (6) is a valid signal for graph structure updates.
    Eq (6) is introduced as a behavioral deviation measure, but its properties and thresholds are not analyzed.

pith-pipeline@v1.4.0-alltime-deepseek-medium · 10233 in / 12687 out tokens · 132924 ms · 2026-08-05T05:43:35.940298+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Topology-Aware Graph Reinforcement Learning for Dynamic Routing in Cloud Networks." pith.science (2026). https://pith.science/paper/TCGYOYDJ

@misc{pith2026250904973,
  author       = {Pith},
  title        = {Pith review of: Topology-Aware Graph Reinforcement Learning for Dynamic Routing in Cloud Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TCGYOYDJ}},
  note         = {Machine review of arXiv:2509.04973}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

This paper proposes a topology-aware graph reinforcement learning approach to address the routing policy optimization problem in cloud server environments. The method builds a unified framework for state representation and structural evolution by integrating a Structure-Aware State Encoding (SASE) module and a Policy-Adaptive Graph Update (PAGU) mechanism. It aims to tackle the challenges of decision instability and insufficient structural awareness under dynamic topologies. The SASE module models node states through multi-layer graph convolution and structural positional embeddings, capturing high-order dependencies in the communication topology and enhancing the expressiveness of state representations. The PAGU module adjusts the graph structure based on policy behavior shifts and reward feedback, enabling adaptive structural updates in dynamic environments. Experiments are conducted on the real-world GEANT topology dataset, where the model is systematically evaluated against several representative baselines in terms of throughput, latency control, and link balance. Additional experiments, including hyperparameter sensitivity, graph sparsity perturbation, and node feature dimensionality variation, further explore the impact of structure modeling and graph updates on model stability and decision quality. Results show that the proposed method outperforms existing graph reinforcement learning models across multiple performance metrics, achieving efficient and robust routing in dynamic and complex cloud networks.

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

56 extracted references · 56 canonical work pages · 10 internal anchors

  1. [1]

    Packet routing with graph attention multi- agent reinforcement learning

    Mai X., Fu Q., Chen Y., "Packet routing with graph attention multi- agent reinforcement learning", Proceedings of the 2021 IEEE Global CommunicationsConference(GLOBECOM),pp.1-6,2021

    work page 2021

  2. [2]

    ENERO: Efficient real-time WAN routing optimization with Deep Reinforcement Learning

    Almasan P., Xiao S., Cheng X., et al., "ENERO: Efficient real-time WAN routing optimization with Deep Reinforcement Learning", ComputerNetworks,vol.214,109166,2022

    work page 2022

  3. [3]

    GRL-PS: Graph embedding-based DRL approach for adaptive path selection

    Wei W., Fu L., Gu H., et al., "GRL-PS: Graph embedding-based DRL approach for adaptive path selection", IEEE Transactions on Network andServiceManagement,vol.20,no.3,pp.2639-2651,2023

    work page 2023

  4. [4]

    Advancing Corporate Financial Forecasting: The Role of LSTM and AI in Modern Accounting

    Bao Q., "Advancing Corporate Financial Forecasting: The Role of LSTM and AI in Modern Accounting", Transactions on Computational andScientificMethods,vol.4,no.6,2024

    work page 2024

  5. [5]

    Optimizedconvolutionalneuralnetworkforintelligent financial statement anomaly detection

    DuX.,"Optimizedconvolutionalneuralnetworkforintelligent financial statement anomaly detection", Journal of Computer Technology and Software,vol.3,no.9,2024

    work page 2024

  6. [6]

    Time Series ForecastingwithAttention-AugmentedRecurrentNetworks:AFinancial MarketApplication

    Xu Z., Liu X., Xu Q., Su X., Guo X., and Wang Y., "Time Series ForecastingwithAttention-AugmentedRecurrentNetworks:AFinancial MarketApplication",2025

    work page 2025

  7. [7]

    Forecasting Asset Returns with Structured Text Factors and Dynamic Time Windows

    Su X., "Forecasting Asset Returns with Structured Text Factors and Dynamic Time Windows", Transactions on Computational and ScientificMethods,vol.4,no.6,2024

    work page 2024

  8. [8]

    BERT-Based Automatic Audit Report Generation and Compliance Analysis

    Xu Z., Sheng Y., Bao Q., Du X., Guo X., and Liu Z., "BERT-Based Automatic Audit Report Generation and Compliance Analysis", Proceedings of the 2025 5th International Conference on Artificial Intelligence and Industrial Technology Applications (AIITA), pp. 1233- 1237,Mar.2025

    work page 2025

  9. [9]

    Semantic and Structural Analysis of Implicit Biases in Large Language Models: An Interpretable Approach

    ZhangR., LianL., QiZ., andLiu G.,"Semantic andStructural Analysis of Implicit Biases in Large Language Models: An Interpretable Approach",arXive-prints,arXiv:2508.06155,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  10. [10]

    Structured Memory Mechanisms for Stable Context Representation in Large Language Models

    Xing Y., Yang T., Qi Y., Wei M., Cheng Y., and Xin H., "Structured Memory Mechanisms for Stable Context Representation in Large LanguageModels",arXive-prints,arXiv:2505.22921,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  11. [11]

    Structured Compression of Large Language Models with Sensitivity-aware Pruning Mechanisms

    Wang Y., "Structured Compression of Large Language Models with Sensitivity-aware Pruning Mechanisms", Journal of Computer TechnologyandSoftware,vol.3,no.9,2024

    work page 2024

  12. [12]

    Internal Knowledge Adaptation in LLMs with Consistency- Constrained Dynamic Routing

    Wu Q., "Internal Knowledge Adaptation in LLMs with Consistency- Constrained Dynamic Routing", Transactions on Computational and ScientificMethods,vol.4,no.5,2024

    work page 2024

  13. [13]

    Target-Oriented Causal Representation Learning for Robust Cross-Market Return Prediction

    Wang Y., Sha Q., Feng H., and Bao Q., "Target-Oriented Causal Representation Learning for Robust Cross-Market Return Prediction", Journal of Computer Science and Software Applications, vol. 5, no. 5, 2025

    work page 2025

  14. [14]

    Enhancing Systemic Risk Forecasting with Deep Attention Models in Financial TimeSeries

    Xu Q. R., Xu W., Su X., Ma K., Sun W., and Qin Y., "Enhancing Systemic Risk Forecasting with Deep Attention Models in Financial TimeSeries",2025

    work page 2025

  15. [15]

    Financial text analysis using 1D-CNN: Risk classification and auditing support

    Du X., "Financial text analysis using 1D-CNN: Risk classification and auditing support", Proceedings of the 2025 International Conference on ArtificialIntelligenceandComputationalIntelligence,pp.515-520, Feb. 2025

    work page 2025

  16. [16]

    Hierarchical semantic-structural encoding for compliance risk detection with LLMs

    Qin Y., "Hierarchical semantic-structural encoding for compliance risk detection with LLMs", Transactions on Computational and Scientific Methods,vol.4,no.6,2024

    work page 2024

  17. [17]

    Pre-trained LanguageModelsandFew-shotLearningforMedicalEntityExtraction

    Wang X., Liu G., Zhu B., He J., Zheng H., and Zhang H., "Pre-trained LanguageModelsandFew-shotLearningforMedicalEntityExtraction", Proceedings of the 2025 5th International Conference on Artificial Intelligence and Industrial Technology Applications (AIITA), pp. 1243- 1247,Mar.2025

    work page 2025

  18. [18]

    Time-Aware and Multi-Source Feature Fusion for Transformer-Based Medical Text Analysis

    Wang X., "Time-Aware and Multi-Source Feature Fusion for Transformer-Based Medical Text Analysis", Transactions on ComputationalandScientificMethods,vol.4,no.7,2024

    work page 2024

  19. [19]

    Structure-Aware Temporal Modeling for Chronic Disease Progression Prediction

    Hu J., Zhang B., Xu T., Yang H., and Gao M., "Structure-Aware Temporal Modeling for Chronic Disease Progression Prediction", arXiv e-prints,arXiv:2508.14942,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  20. [20]

    Joint Modeling of Medical Images and Clinical Text for Early Diabetes Risk Detection

    Zi Y. and Deng X., "Joint Modeling of Medical Images and Clinical Text for Early Diabetes Risk Detection", Journal of Computer TechnologyandSoftware,vol.4,no.7,2025

    work page 2025

  21. [21]

    Deep Skin Lesion Segmentation with Transformer-CNN Fusion: Toward Intelligent Skin Cancer Analysis

    Wang X., Zhang X., and Wang X., "Deep Skin Lesion Segmentation with Transformer-CNN Fusion: Toward Intelligent Skin Cancer Analysis",arXive-prints,arXiv:2508.14509,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  22. [22]

    Sequential Recommendation via Time- Aware and Multi-Channel Convolutional User Modeling

    Xing Y., Wang Y., and Zhu L., "Sequential Recommendation via Time- Aware and Multi-Channel Convolutional User Modeling", Transactions onComputationalandScientificMethods,vol.5,no.5,2025

    work page 2025

  23. [23]

    Analyzing data augmentation techniques for contrastive learning in recommender models

    Wei M., Xin H., Qi Y., Xing Y., Ren Y., and Yang T., "Analyzing data augmentation techniques for contrastive learning in recommender models",2025

    work page 2025

  24. [24]

    Knowledge Graph-Infused Fine-Tuning for Structured Reasoning in Large Language Models

    Zhang W.,Tian Y., Meng X., Wang M., andDu J., "Knowledge Graph- Infused Fine-Tuning for Structured Reasoning in Large Language Models",arXive-prints,arXiv:2508.14427,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  25. [25]

    Layer-Wise Structural Mapping for Efficient Domain Transfer in Language Model Distillation

    Quan X., "Layer-Wise Structural Mapping for Efficient Domain Transfer in Language Model Distillation", Transactions on ComputationalandScientificMethods,vol.4,no.5,2024

    work page 2024

  26. [26]

    Selective Knowledge Injection via Adapter Modules in Large ‐Scale Language Models

    Zheng H., Zhu L., Cui W., Pan R., Yan X., and Xing Y., "Selective Knowledge Injection via Adapter Modules in Large ‐Scale Language Models",2025

    work page 2025

  27. [27]

    GROM: A generalized routing optimization method with graph neural network and deep reinforcement learning

    Ding M., Guo Y., Huang Z., et al., "GROM: A generalized routing optimization method with graph neural network and deep reinforcement learning", Journal of Network and Computer Applications, vol. 229, 103927,2024

    work page 2024

  28. [28]

    Unsupervised Temporal Encoding for Stock Price Prediction throughDual-PhaseLearning

    Xu Q., "Unsupervised Temporal Encoding for Stock Price Prediction throughDual-PhaseLearning",2025

    work page 2025

  29. [29]

    Hybrid Deep Learning for Financial Volatility Forecasting: An LSTM-CNN-Transformer Model

    Sha Q., "Hybrid Deep Learning for Financial Volatility Forecasting: An LSTM-CNN-Transformer Model", Transactions on Computational and ScientificMethods,vol.4,no.11,2024

    work page 2024

  30. [30]

    Graphreinforcementlearningformulti- aircraft conflict resolution

    LiY.,ZhangY.,GuoT.,etal.,"Graphreinforcementlearningformulti- aircraft conflict resolution", IEEE Transactions on Intelligent Vehicles, vol.9,no.3,pp.4529-4540,2024

    work page 2024

  31. [31]

    Graph Reinforcement Learning in Power Grids: A Survey

    Hassouna M., Holzhüter C., Lytaev P., et al., "Graph Reinforcement Learning in Power Grids: A Survey", arXiv e-prints, arXiv:2407.04522, 2024

    work page arXiv 2024

  32. [32]

    Graph Reinforcement and Asynchronous Federated Learning based task offloading in fog computing

    Verma N. K., Naik K. J., "Graph Reinforcement and Asynchronous Federated Learning based task offloading in fog computing", Physical Communication,102776,2025

    work page 2025

  33. [33]

    Learning and generating distributed routing protocols using graph-based deep learning

    Geyer F., Carle G., "Learning and generating distributed routing protocols using graph-based deep learning", Proceedings of the 2018 Workshop on Big Data Analytics and Machine Learning for Data CommunicationNetworks,pp.40-45,2018

    work page 2018

  34. [34]

    Strategic Cache Allocation via Game-Aware Multi-Agent ReinforcementLearning

    Ren Y., "Strategic Cache Allocation via Game-Aware Multi-Agent ReinforcementLearning", Transactions on Computational and Scientific Methods,vol.4,no.8,2024

    work page 2024

  35. [35]

    Graph Neural Network-Based Collaborative Perception for Adaptive Scheduling in Distributed Systems

    ZhuW.,WuQ.,TangT.,MengR.,ChaiS.,andQuanX.,"GraphNeural Network-Based Collaborative Perception for Adaptive Scheduling in DistributedSystems",arXive-prints,arXiv:2505.16248,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  36. [36]

    Entity-Aware Graph Neural Modeling for Structured Information Extraction in the Financial Domain

    Wang Y., "Entity-Aware Graph Neural Modeling for Structured Information Extraction in the Financial Domain", Transactions on ComputationalandScientificMethods,vol.4,no.9,2024

    work page 2024

  37. [37]

    Selective Noise Injection and Feature Scoring for Unsupervised Request Anomaly Detection

    Cheng Y., "Selective Noise Injection and Feature Scoring for Unsupervised Request Anomaly Detection", Journal of Computer TechnologyandSoftware,vol.3,no.9,2024

    work page 2024

  38. [38]

    Privacy-Enhanced FederatedLearningforDistributedHeterogeneousData

    Kang T., Yang H., Dai L., Hu X., and Du J., "Privacy-Enhanced FederatedLearningforDistributedHeterogeneousData",2025

    work page 2025

  39. [39]

    AI-Driven Multi-Agent Scheduling and Service Quality Optimization in Microservice Systems

    Zhang R., "AI-Driven Multi-Agent Scheduling and Service Quality Optimization in Microservice Systems", Transactions on Computational andScientificMethods,vol.5,no.8,2025

    work page 2025

  40. [40]

    Graph Neural Network and Transformer Integration for Unsupervised System Anomaly Discovery

    Zi Y., Gong M., Xue Z., Zou Y., Qi N., and Deng Y., "Graph Neural Network and Transformer Integration for Unsupervised System AnomalyDiscovery",arXive-prints,arXiv:2508.09401,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  41. [41]

    EEG Anomaly Detection Using Temporal Graph Attention for Clinical Applications

    Zhang X. and Wang Q., "EEG Anomaly Detection Using Temporal Graph Attention for Clinical Applications", Journal of Computer TechnologyandSoftware,vol.4,no.7,2025

    work page 2025

  42. [42]

    Application of Deep Generative Models for Anomaly Detection in Complex Financial Transactions

    Tang T., Yao J., Wang Y., Sha Q., Feng H., and Xu Z., "Application of Deep Generative Models for Anomaly Detection in Complex Financial Transactions", Proceedings of the 2025 4th International Conference on Artificial Intelligence, Internet and Digital Economy (ICAID), pp. 133- 137,Apr.2025

    work page 2025

  43. [43]

    Single-Device Human Activity Recognition Based on Spatiotemporal Feature Learning Networks

    Zhan J., "Single-Device Human Activity Recognition Based on Spatiotemporal Feature Learning Networks", Transactions on ComputationalandScientificMethods,vol.5,no.3,2025

    work page 2025

  44. [44]

    Evaluating explainability forgraphneuralnetworks

    Agarwal C., Queen O., Lakkaraju H., et al., "Evaluating explainability forgraphneuralnetworks",ScientificData,vol.10,no.1,144,2023

    work page 2023

  45. [45]

    GraphNeuralRecognitionofMaliciousUserPatternsinCloud Systems via Attention Optimization

    GaoD.,"GraphNeuralRecognitionofMaliciousUserPatternsinCloud Systems via Attention Optimization", Transactions on Computational andScientificMethods,vol.4,no.12,2024

    work page 2024

  46. [46]

    Behavioral Anomaly Detection in Distributed Systems via Federated Contrastive Learning

    Meng R., Wang H., Sun Y., Wu Q., Lian L., and Zhang R., "Behavioral Anomaly Detection in Distributed Systems via Federated Contrastive Learning",arXive-prints,arXiv:2506.19246,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  47. [47]

    OptimizingDistributedComputingResourceswithFederated Learning: Task Scheduling and Communication Efficiency

    WangY.,"OptimizingDistributedComputingResourceswithFederated Learning: Task Scheduling and Communication Efficiency", Journal of ComputerTechnologyandSoftware,vol.4,no.3,2025

    work page 2025

  48. [48]

    A Deep LearningFrameworkforSequenceMiningwithBidirectionalLSTMand Multi-Scale Attention

    Yang T., Cheng Y., Ren Y., Lou Y., Wei M., and Xin H., "A Deep LearningFrameworkforSequenceMiningwithBidirectionalLSTMand Multi-Scale Attention", Proceedings of the 2025 2nd International Conference on Innovation Management and Information System, pp. 472-476,Apr.2025

    work page 2025

  49. [49]

    Causal Discriminative Modeling for Robust Cloud Service Fault Detection

    Wang H., "Causal Discriminative Modeling for Robust Cloud Service Fault Detection",JournalofComputer TechnologyandSoftware, vol.3, no.7,2024

    work page 2024

  50. [50]

    Domain-Adversarial Transfer Learning for Fault Root Cause Identification in Cloud Computing Systems

    Fang B. and Gao D., "Domain-Adversarial Transfer Learning for Fault RootCauseIdentificationinCloudComputingSystems",arXive-prints, arXiv:2507.02233,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  51. [51]

    ScalableMulti- PartyCollaborativeDataMiningBasedonFederatedLearning

    WangM.,KangT.,DaiL.,YangH.,DuJ.,andLiuC.,"ScalableMulti- PartyCollaborativeDataMiningBasedonFederatedLearning",2025

    work page 2025

  52. [52]

    Artificial Intelligence-Based Multiscale Temporal Modeling for Anomaly Detection in Cloud Services

    Lian L., Li Y., Han S., Meng R., Wang S., and Wang M., "Artificial Intelligence-Based Multiscale Temporal Modeling for Anomaly DetectioninCloudServices",arXive-prints,arXiv:2508.14503,2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  53. [53]

    GDDR: GNN-based data-driven routing

    Hope O., Yoneki E., "GDDR: GNN-based data-driven routing", Proceedings of the 2021 IEEE 41st International Conference on DistributedComputingSystems(ICDCS),pp.517-527,2021

    work page 2021

  54. [54]

    DeepCQ+: Robust and scalable routingwithmulti-agentdeepreinforcementlearningforhighlydynamic networks

    Kaviani S., Ryu B., Ahmed E., et al., "DeepCQ+: Robust and scalable routingwithmulti-agentdeepreinforcementlearningforhighlydynamic networks", Proceedings of the MILCOM 2021-2021 IEEE Military CommunicationsConference(MILCOM),pp.31-36,2021

    work page 2021

  55. [55]

    CFR-RL: Traffic engineering with reinforcement learning in SDN

    Zhang J., Ye M., Guo Z., et al., "CFR-RL: Traffic engineering with reinforcement learning in SDN", IEEE Journal on Selected Areas in Communications,vol.38,no.10,pp.2249-2259,2020

    work page 2020

  56. [56]

    Graph-based reinforcement learning for software-defined networking traffic engineering

    Lu J., Tang C., Ma W., et al., "Graph-based reinforcement learning for software-defined networking traffic engineering", Journal of King Saud UniversityComputerandInformationSciences,vol.37,no.6,119,2025

    work page 2025