Pith. sign in

REVIEW 4 major objections 5 minor 69 references

By encoding travel demand as scalar attributes on virtual links, GUIDED decouples input dimension from network size, making GNN-based traffic surrogates transferable across urban topologies without zero-padding.

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

2026-08-01 12:52 UTC pith:T3Y4RKRE

load-bearing objection Nice idea — scalar demand on virtual edges pooled to node embeddings — but the evaluation confounds compactness with the GUIDED mechanism, and the ablation undercuts the attribution; still worth a referee. the 4 major comments →

arxiv 2607.19270 v1 pith:T3Y4RKRE submitted 2026-07-21 cs.LG cs.AI

GUIDED Network-Agnostic Feature Initialization for Spatial Transferability in GNN-based Models

Alessandro Scalese , Santhanakrishnan Narayanan , Constantinos Antoniou This is my paper
classification cs.LG cs.AI
keywords traffic assignment problemgraph neural networksnetwork-agnostic initializationspatial generalizationtransfer learningvirtual linksheterogeneous graphsdomain adaptation
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.

The paper claims that the spatial generalization gap in GNN-based surrogates for the traffic assignment problem can be closed by changing where demand enters the network. It proposes GUIDED, an initialization layer that places each origin-destination demand as a scalar attribute on an auxiliary virtual link and aggregates these edge embeddings into node features. Because the input dimension no longer depends on network size, the same model can be applied to different cities without zero-padding. The authors report that the approach maintains state-of-the-art accuracy, improves robustness to out-of-distribution demand, halves per-epoch training time, and enables parameter-efficient transfer learning.

Core claim

The central claim is that feature initialization, not message passing, is the root of the transductive limitation in GNN traffic surrogates. GUIDED replaces a node feature matrix whose columns correspond to network origins with a two-stage construction: scalar demands are embedded on virtual OD edges, then a permutation-invariant sum over incoming and outgoing virtual edges forms each node's initial embedding. This keeps the latent dimension fixed at 64 regardless of how many nodes or OD pairs the network has. The authors show that GUIDED integrated into a heterogeneous graph attention architecture matches or exceeds the baseline on single-network tasks, and that fine-tuning only the last re

What carries the argument

The key mechanism is the virtual-link embedding plus aggregation scheme. Each nonzero origin-destination demand is attached as a scalar to a directed virtual edge connecting the origin and destination nodes. A projection (either a two-layer MLP with a 16-unit hidden layer, or a 20-centered RBF expansion followed by a two-layer MLP) lifts that scalar into a 32-dimensional edge embedding. Each node then sums the embeddings of its outgoing and incoming virtual edges separately and concatenates the two sums, producing a fixed 64-dimensional node embedding that is independent of network size. This input standardization is what makes downstream message passing network-agnostic.

Load-bearing premise

The paper attributes the transfer gains to the virtual-edge aggregation, but its own ablation shows that adapting the initialization layer adds little beyond adapting the last routing and prediction layers, leaving open the possibility that the compact 64-dimensional input alone is driving the improvement.

What would settle it

Train the baseline heterogeneous graph attention model with a fixed 64-dimensional node feature obtained by a random projection (or a learned matrix) of the OD matrix, keeping all other settings identical to the GUIDED experiments. If this modified baseline matches GUIDED's transfer performance, then the virtual-edge aggregation mechanism is not the cause of the reported gains; alternatively, if it falls short, the aggregation is carrying the effect.

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

If this is right

  • A GNN surrogate trained on one city's network can be transferred to another city with only a small subset of parameters fine-tuned, eliminating the need for input padding or architectural changes.
  • Because input dimension no longer grows with the number of origins, the same model definition can serve networks of arbitrary scale, provided memory for virtual edges is available.
  • The reported roughly 50% reduction in per-epoch training time from scatter-based aggregation makes large-scale surrogate training cheaper.
  • The framework generalizes the virtual-link paradigm from node features to edge features, and the authors argue this extends to any origin-destination spatial interaction problem such as freight or multimodal networks.

Where Pith is reading between the lines

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

  • The paper's own layer-wise ablation shows that unfreezing the GUIDED initialization parameters contributes little; the decisive adapted parameters are the final real-edge encoder layer and the predictor. This suggests the performance advantage may stem primarily from the compact, fixed 64-dimensional input representation rather than from the virtual-edge aggregation specifically.
  • A direct test would compare GUIDED against a baseline that uses a fixed-size random projection of the OD matrix to 64 dimensions. If the random-projection baseline matches GUIDED on transfer tasks, then the aggregation mechanism is not the operative ingredient.
  • The out-of-distribution robustness gains might also be explained by the lower-dimensional, denser feature representation reducing overfitting to network-specific node layouts, rather than by any physical inductive bias in the virtual edges.
  • If the mechanism is the compact embedding, a simpler and cheaper solution than virtual edges might exist; if the mechanism is the aggregation, pruning low-demand virtual edges should be possible without accuracy loss, which would address the O(|Z|^2) memory scaling limitation.

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

4 major / 5 minor

Summary. The paper proposes GUIDED, a network-agnostic input-initialization layer for GNN-based traffic assignment surrogates. Instead of placing the full OD matrix on nodes (whose dimension depends on the number of TAZs), it places scalar demands on virtual links and aggregates learned virtual-edge embeddings into fixed-size node embeddings via permutation-invariant sums (Eqs. 4–6). This, the authors argue, decouples input dimensionality from network size, enabling cross-network fine-tuning without zero-padding. The method is integrated into a HetGAT architecture (GUIDED-HetGAT, with linear and RBF variants) and evaluated on two real-world networks (Anaheim, Chicago) under four experiments: intra-network accuracy, data scarcity, semi-supervised partial observability, and inter-network transfer with layer-wise ablation. The paper reports consistent predictive improvements over a HetGAT baseline, better physical consistency (FCN), and roughly 50% lower per-epoch training time, while noting that zero-shot transfer fails and fine-tuning is required.

Significance. If the core claim is established, the GUIDED layer is a simple, modular contribution: it replaces a transductive OD-matrix node feature layout with scalar edge attributes plus a permutation-invariant aggregation, yielding a fixed-dimensional input without zero-padding. The experimental design is broader than many prior GNN-TAP surrogates: two networks, two demand-generation regimes, four tasks, and a layer-wise transfer ablation. The authors also provide the full codebase (on request), document the UE ground-truth generation (BFW, rgap=1e-5), and report domain-specific metrics (%GEH<5, FCN), which are appropriate for the field. The main weaknesses are that the central attribution of gains to the GUIDED mechanism is not isolated from input-dimensionality differences, no error bars or significance testing are reported, and the 'state-of-the-art' claim is made without comparing to other recent TAP-GNN methods. These are fixable with additional experiments and careful language.

major comments (4)
  1. [§4.2.1, §5.4, Tables 5-6] The reported gains are confounded by input dimensionality. The baseline HetGAT consumes a zero-padded 950-dimensional OD vector even for Anaheim (416 nodes), while GUIDED-HetGAT consumes a 64-dimensional dense embedding. The paper's own ablation does not resolve this: unfreezing Θ_preproc in addition to Θ_RL and Θ_pred gives non-negligible improvements in multiple cases (e.g., AN-A Dataset A lin: 25.22→20.86; AN-B Dataset B baseline: 126.60→110.54 vs lin: 104.69→96.35). The interpretation that the initialization layer is not critical (Section 5.4) is not uniformly supported by the table. A controlled baseline with an equivalent 64-dimensional dense input (e.g., learned projection of the OD matrix without virtual-edge aggregation, or a 64-dimensional row/column embedding) is required to isolate the contribution of the virtual-edge aggregation mechanism.
  2. [Tables 4-6] All results are reported as single-point estimates with no error bars, multiple seeds, or significance tests. Many claimed improvements are small (e.g., CH-A Experiment A: MAE 14.09 vs 13.38; CH-A Experiment B: %GEH 98.29 vs 98.12), and without run-to-run variance it is impossible to judge whether the consistent direction of the tables reflects a real effect or noise. At minimum, the headline experiments should be run with 3–5 seeds and reported as mean ± std, with a paired significance test where possible.
  3. [Abstract, §5.1] The term 'state-of-the-art predictive accuracy' is used without comparison to any method other than the HetGAT baseline. The literature review lists several recent alternatives for GNN-based traffic assignment (e.g., Lassen et al., 2025; Ameli et al., 2026; Hu & Xie, 2025), but none is benchmarked. Either add quantitative comparisons to at least one of these recent methods, or qualify the claim to 'competitive with the HetGAT baseline' and 'state-of-the-art among the tested configurations.'
  4. [§5.4, §6] Zero-shot transfer fails catastrophically for both the baseline and the proposed model, with negative R² values (e.g., Table 5, AN-A HetGAT R²=-2.201; lin R²=-0.050). This undercuts the abstract/conclusion language that GUIDED 'renders the downstream architecture completely agnostic to the size of the network' and enables 'seamless transitions between networks.' The mechanism only helps after fine-tuning. The paper should state this limitation explicitly in the abstract and frame 'network-agnostic' as applying to input representation, not to zero-shot cross-network predictive capability.
minor comments (5)
  1. [Tables 4-6] Arrow directions for RMSN and %GEH<5 are inconsistent: Table 4 shows RMSN↑ and %GEH<5↓, while Tables 5 and 6 show RMSN↓ and %GEH<5↑. RMSN is normally a lower-is-better metric; please standardize.
  2. [§6] Typo: 'This ia structural limitation' should read 'This is a structural limitation.'
  3. [§4.2.1] The zero-padding of the baseline to 950 nodes is applied even in single-network tasks (Experiments A-C). Please state explicitly whether the baseline is zero-padded for those experiments as well, and justify why this is not a handicap for the baseline. This is related to the major confound but should at least be transparent in the setup.
  4. [§3.2, §4.2.2] The hyperparameters of the GUIDED layer (RBF centers, bandwidth, number of centers K=20, loss weights λ_v=1.0, λ_f=0.003, λ_c=0.003) are described as 'empirically determined' or 'manually tuned,' but no sensitivity analysis is provided. Since the loss weights directly affect the comparison between methods, a short sensitivity study or a note that results are robust to reasonable variations would strengthen the paper.
  5. [§4.2.1] The code repository is private and 'available upon request.' For reproducibility, especially in a methods paper, the code and data-generation scripts should be made publicly available in the final version.

Circularity Check

0 steps flagged

No significant circularity: the paper's predictions are evaluated against external UE ground truth and the proposed initialization is not defined in terms of the target quantities.

full rationale

The paper does not contain a derivation chain that reduces to its inputs. The central claim is that replacing OD-matrix node features with scalar demand attributes on virtual links, followed by a learned aggregation into fixed-dimension node embeddings, enables network-agnostic transfer. This is an architectural proposal whose output is compared against ground-truth link flows computed by the external BFW solver (Section 4.1.2, 'Ground Truth Generation: UE Solutions'), not against the model's own fitted values. The GUIDED layer is a learned feature initialization, not a fitted constant renamed as a prediction. The linear and RBF projection functions are explicit (Eqs. 1-3) and the aggregation is explicit (Eqs. 4-6); no equation defines the output quantity in terms of the prediction target. The loss function is a standard weighted composite (Eqs. 7-10), with manually tuned weights, but these weights do not define or predetermine the evaluated metrics. Self-citations (Narayanan et al. 2024; Makarov et al. 2024) are contextual literature citations and are not load-bearing for the paper's novelty claim. The virtual-link paradigm is attributed to external prior work (Liu & Meidani 2024), which strengthens rather than weakens the independence of the setup. The paper's own ablation (Experiment D, Tables 5-6) shows that unfreezing the initialization layer in addition to the last R-Encoder and predictor yields only modest gains, and the paper itself states that 'the bottleneck in transfer learning performance does not lie in the demand projection itself.' This weakens the attribution of gains to the GUIDED aggregation mechanism, but that is a confounding/ablation concern, not circularity: it does not mean the GUIDED output is equivalent by construction to the prediction target or to a fitted parameter. The zero-padded baseline and compact fixed-dimension embedding differ in input dimensionality, so the comparison may be imperfectly controlled, but again this is an experimental-design limitation, not a definitional reduction. Overall, the empirical predictions are externally grounded, the proposed layer is a genuine architectural component, and no step in the paper's reasoning exhibits self-definitional circularity, fitted-input-called-prediction, or load-bearing self-citation.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

The central claim rests on the BFW ground truth, the synthetic demand generation processes (with hand-set α), and the specific GUIDED aggregation design. No new physical entities are introduced; virtual links are inherited from prior work.

free parameters (4)
  • RBF centers μ_k (K=20)
    Equation (2) uses predefined centers for the RBF expansion; the paper does not state how they are chosen, only that they are 'a set of K centers'.
  • Loss weights λ_v, λ_f, λ_c = 1.0, 0.003, 0.003
    Set via manual tuning (Section 4.2.2) to balance V/C, flow, and conservation loss terms.
  • Dirichlet concentration α = 0.05
    Chosen empirically to make the synthetic Dataset B resemble the original demand in sparsity and skewness (Section 4.1.2); it controls the difficulty of the stress-test dataset.
  • Node embedding dimension F = 64
    Set to match the baseline HetGAT latent dimension (Section 3.3.1); not derived from any principle.
axioms (4)
  • domain assumption BFW algorithm with rgap 10^-5 computes ground-truth UE solutions
    All training labels are generated by this solver (Section 4.1.2); the model's accuracy is measured against these solutions.
  • domain assumption Wardrop's first principle and static demand adequately represent traffic
    The model learns to approximate user equilibrium; the paper itself acknowledges in Section 6 that this ignores temporal dynamics, multi-modal interactions, and bounded rationality.
  • ad hoc to paper The sum-aggregation (Eqs. 4-6) preserves demand information sufficient for flow prediction
    The specific permutation-invariant summation is a design choice; no theoretical guarantee is offered that this aggregation is informative or optimal.
  • standard math Min-Max normalization and train/val/test split prevent data leakage
    Section 4.2.1 computes normalization parameters strictly from the training subset, following standard practice.

pith-pipeline@v1.3.0-alltime-deepseek · 26954 in / 12610 out tokens · 109810 ms · 2026-08-01T12:52:46.344009+00:00 · methodology

0 comments
read the original abstract

The Traffic Assignment Problem is a fundamental but computationally expensive component of transportation planning. While Graph Neural Networks have emerged as fast, data-driven surrogates, their practical deployment is severely constrained by a spatial generalization gap. Standard models rely on transductive feature initializations that tie travel demand to fixed network topologies, preventing seamless transfer to new urban environments. To overcome this structural limitation, this research proposes a network-agnostic initialization layer, termed Geometrically Unconstrained Inductive Demand EmbeDding (GUIDED). By injecting travel demand as a scalar attribute on auxiliary virtual links rather than as specific node features, this modular framework standardizes the input space regardless of network scale. Extensive experimental evaluation across multiple urban topologies demonstrates that a Heterogeneous Graph Attention Network (HetGAT) model integrated with the proposed GUIDED layer maintains state-of-the-art predictive accuracy on single-network tasks, while demonstrating superior robustness to out-of-distribution demand patterns and maintaining a distinct performance advantage over the baseline even under severe data scarcity. Notably, the proposed feature initialization enables highly parameter-efficient domain adaptation for inter-network transfer learning without artificial input homogenization, establishing a robust foundation for truly inductive models. At the same time, the optimized scatter operations of the initialization layer yield an approximate 50% reduction in training time per epoch compared to the baseline approach. Furthermore, while demonstrated on vehicular traffic, this fundamental abstraction of spatial topology provides a versatile blueprint for generalized origin-destination spatial problems, such as freight logistics and multimodal network optimization.

Figures

Figures reproduced from arXiv: 2607.19270 by Alessandro Scalese, Constantinos Antoniou, Santhanakrishnan Narayanan.

Figure 1
Figure 1. Figure 1: Overview of the research structure and methodological framework, mapping the top-down progression of the study. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Visual representation of the Anaheim (left) and Chicago Sketch (right) network topologies. [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of the spatial distribution of the original demand matrix for the Anaheim network (left) with two random [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of the distribution of link capacities and equilibrium flows across scenarios in Datasets A (Random [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of the computational efficiency of the proposed GUIDED-HetGAT models against the baseline HetGAT [PITH_FULL_IMAGE:figures/full_fig_p024_5.png] view at source ↗

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

69 extracted references · 12 linked inside Pith

  1. [1]

    arXiv preprint arXiv:2603.06763 , eprint =

    Metalearning Traffic Assignment for Network Disruptions with Graph Convolutional Neural Networks , author =. arXiv preprint arXiv:2603.06763 , eprint =

    arXiv

  2. [2]

    arXiv preprint arXiv:2006.05205 , eprint =

    On the Bottleneck of Graph Neural Networks and Its Practical Implications , author =. arXiv preprint arXiv:2006.05205 , eprint =

    Pith/arXiv arXiv 2006

  3. [3]

    Ansel, Jason and Yang, Edward and He, Horace and Gimelshein, Natalia and Jain, Animesh and Voznesensky, Michael and Bao, Bin and Bell, Peter and Berard, David and Burovski, Evgeni and Chauhan, Geeta and Chourdia, Anjali and Constable, Will and Desmaison, Alban and DeVito, Zachary and Ellison, Elias and Feng, Will and Gong, Jiong and Gschwind, Michael and ...

    arXiv

  4. [4]

    arXiv preprint arXiv:1607.06450 , eprint =

    Layer Normalization , author =. arXiv preprint arXiv:1607.06450 , eprint =

    Pith/arXiv arXiv

  5. [5]

    Transportation Science , volume =

    Origin-Based Algorithm for the Traffic Assignment Problem , author =. Transportation Science , volume =

  6. [6]

    Studies in the Economics of Transportation , author =

  7. [7]

    Discrete Choice Analysis: Theory and Application to Travel Demand , author =

  8. [8]

    Pattern Recognition and Machine Learning , author =

  9. [9]

    arXiv preprint arXiv:2508.02723 , eprint =

    Mathematical Foundations of Geometric Deep Learning , author =. arXiv preprint arXiv:2508.02723 , eprint =

    Pith/arXiv arXiv

  10. [10]

    Journal of Transportation Engineering , volume =

    Convergence of Traffic Assignments: How Much Is Enough? , author =. Journal of Transportation Engineering , volume =

  11. [11]

    arXiv preprint arXiv:1711.07553 , eprint =

    Residual Gated Graph Convnets , author =. arXiv preprint arXiv:1711.07553 , eprint =

    Pith/arXiv arXiv

  12. [12]

    arXiv preprint arXiv:2105.14491 , eprint =

    How Attentive Are Graph Attention Networks? , author =. arXiv preprint arXiv:2105.14491 , eprint =

    Pith/arXiv arXiv

  13. [13]

    arXiv preprint arXiv:1312.6203 , eprint =

    Spectral Networks and Locally Connected Networks on Graphs , author =. arXiv preprint arXiv:1312.6203 , eprint =

    Pith/arXiv arXiv

  14. [14]

    Encyclopedia of Operations Research and Management Science

    Wardrop Equilibria , author =. Encyclopedia of Operations Research and Management Science. Wiley , keywords =

  15. [15]

    Mathematics of control, signals and systems , volume =

    Approximation by Superpositions of a Sigmoidal Function , author =. Mathematics of control, signals and systems , volume =

  16. [16]

    Modelling Transport , author =

  17. [17]

    Advances in neural information processing systems , volume =

    Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering , author =. Advances in neural information processing systems , volume =

  18. [18]

    Transportation Research Part B: Methodological , volume =

    A Path-Based User-Equilibrium Traffic Assignment Algorithm That Obviates Path Storage and Enumeration , author =. Transportation Research Part B: Methodological , volume =

  19. [19]

    Traffic Analysis Toolbox, Volume

    Dowling, Richard and Skabardonis, Alexander and Alexiadis, Vassili , year = 2004, keywords =. Traffic Analysis Toolbox, Volume

    2004

  20. [20]

    arXiv preprint arXiv:2012.09699 , eprint =

    A Generalization of Transformer Networks to Graphs , author =. arXiv preprint arXiv:2012.09699 , eprint =

    Pith/arXiv arXiv 2012

  21. [21]

    City-Agnostic Demand Prediction: A Graph Attention Approach for Urban Transfer Learning , booktitle =

    Erdol, Hakan and Klar, Robert and Angelakis, Vangelis and Pope, James and Piechocki, Robert and Tryfonas, Theo and Oikonomou, George , year = 2025, pages =. City-Agnostic Demand Prediction: A Graph Attention Approach for Urban Transfer Learning , booktitle =

    2025

  22. [22]

    The Gravity Model in Transportation Analysis: Theory and Extensions , author =

  23. [23]

    Temporal Graph Learning Workshop @ KDD , keywords =

    Fey, Matthias and Sunil, Jinu and Nitta, Akihiro and Puri, Rishi and Shah, Manan and z. Temporal Graph Learning Workshop @ KDD , keywords =

  24. [24]

    Transport reviews , volume =

    How Common and How Large Are Cost Overruns in Transport Infrastructure Projects? , author =. Transport reviews , volume =

  25. [25]

    How (in) Accurate Are Demand Forecasts in Public Works Projects?:

    Flyvbjerg, Bent and Skamris Holm, Mette K and Buhl, S. How (in) Accurate Are Demand Forecasts in Public Works Projects?:. Journal of the American planning association , volume =

  26. [26]

    Naval research logistics quarterly , volume =

    An Algorithm for Quadratic Programming , author =. Naval research logistics quarterly , volume =

  27. [27]

    A Modified

    Fukushima, Masao , year = 1984, journal =. A Modified

    1984

  28. [28]

    Neural Message Passing for Quantum Chemistry , booktitle =

    Gilmer, Justin and Schoenholz, Samuel S and Riley, Patrick F and Vinyals, Oriol and Dahl, George E , year = 2017, pages =. Neural Message Passing for Quantum Chemistry , booktitle =

    2017

  29. [29]

    Message Passing Neural Networks , booktitle =

    Gilmer, Justin and Schoenholz, Samuel S and Riley, Patrick F and Vinyals, Oriol and Dahl, George E , year = 2020, pages =. Message Passing Neural Networks , booktitle =

    2020

  30. [30]

    Deep Learning , author =

  31. [31]

    A New Model for Learning in Graph Domains , booktitle =

    Gori, Marco and Monfardini, Gabriele and Scarselli, Franco , year = 2005, volume =. A New Model for Learning in Graph Domains , booktitle =

    2005

  32. [32]

    Advances in neural information processing systems , volume =

    Implicit Graph Neural Networks , author =. Advances in neural information processing systems , volume =

  33. [33]

    Advances in neural information processing systems , volume =

    Inductive Representation Learning on Large Graphs , author =. Advances in neural information processing systems , volume =

  34. [34]

    Neural networks , volume =

    Approximation Capabilities of Multilayer Feedforward Networks , author =. Neural networks , volume =

  35. [35]

    Measurement Science and Technology , volume =

    A Deep Graph Convolution Spatial-Temporal Attention Learning Model for Traffic Flow Prediction , author =. Measurement Science and Technology , volume =

  36. [36]

    Graphbridge:

    Ju, Li and Yang, Xingyi and Li, Qi and Wang, Xinchao , year = 2025, journal =. Graphbridge:. 2502.19252 , archiveprefix =

    Pith/arXiv arXiv 2025

  37. [37]

    Kingma, Diederik P and Ba, Jimmy , year = 2014, journal =. Adam:. 1412.6980 , archiveprefix =

    Pith/arXiv arXiv 2014

  38. [38]

    arXiv preprint arXiv:1609.02907 , eprint =

    Semi-Supervised Classification with Graph Convolutional Networks , author =. arXiv preprint arXiv:1609.02907 , eprint =

    Pith/arXiv arXiv

  39. [39]

    Learning Traffic Flows:

    Lassen, Oskar Bohn and Agriesti, Serio and Eldafrawi, Mohamed and Gammelli, Daniele and Cantelmo, Guido and Gentile, Guido and Pereira, Francisco Camara , year = 2025, pages =. Learning Traffic Flows:. 2025 9th International Conference on Models and Technologies for Intelligent Transportation Systems (

    2025

  40. [40]

    Transportation Research Part C: Emerging Technologies , volume =

    End-to-End Heterogeneous Graph Neural Networks for Traffic Assignment , author =. Transportation Research Part C: Emerging Technologies , volume =

  41. [41]

    End-to-End Learning of User Equilibrium:

    Liu, Zhichen and Yin, Yafeng , year = 2025, journal =. End-to-End Learning of User Equilibrium:

    2025

  42. [42]

    Expert Systems with Applications , volume =

    Multi-Class Traffic Assignment Using Multi-View Heterogeneous Graph Attention Networks , author =. Expert Systems with Applications , volume =

  43. [43]

    Electronic Research Archive , volume =

    A Scalable Learning Approach for User Equilibrium Traffic Assignment Problem Using Graph Convolutional Networks , author =. Electronic Research Archive , volume =

  44. [44]

    IEEE Transactions on Intelligent Transportation Systems , publisher =

    Spatiotemporal Generalization Graph Neural Network-Based Prediction Models by Considering Morphological Diversity in Traffic Networks , author =. IEEE Transactions on Intelligent Transportation Systems , publisher =

  45. [45]

    Rectifier Nonlinearities Improve Neural Network Acoustic Models , booktitle =

    Maas, Andrew L and Hannun, Awni Y and Ng, Andrew Y and others , year = 2013, volume =. Rectifier Nonlinearities Improve Neural Network Acoustic Models , booktitle =

    2013

  46. [46]

    arXiv preprint arXiv:2408.07726 , eprint =

    Development of a Graph Neural Network Surrogate for Travel Demand Modelling , author =. arXiv preprint arXiv:2408.07726 , eprint =

    Pith/arXiv arXiv

  47. [47]

    Modelling Reservation-Based Shared Autonomous Vehicle Services:

    Narayanan, Santhanakrishnan and Chaniotakis, Emmanouil and Antoniou, Constantinos , year = 2022, journal =. Modelling Reservation-Based Shared Autonomous Vehicle Services:. doi:10.1016/j.trc.2022.103651 , abstract =

    arXiv 2022

  48. [48]

    Graph Neural Networks as Strategic Transport Modelling Alternative-

    Narayanan, Santhanakrishnan and Makarov, Nikita and Antoniou, Constantinos , year = 2024, journal =. Graph Neural Networks as Strategic Transport Modelling Alternative-

    2024

  49. [49]

    Transportation Land Use Modeling and Policy , author =

  50. [50]

    The Traffic Assignment Problem: Models and Methods , author =

  51. [51]

    Data-Driven Traffic Assignment:

    Rahman, Rezaur and Hasan, Samiul , year = 2023, journal =. Data-Driven Traffic Assignment:

    2023

  52. [52]

    Graph Neural Networks for Intelligent Transportation Systems:

    Rahmani, Saeed and Baghbani, Asiye and Bouguila, Nizar and Patterson, Zachary , year = 2023, journal =. Graph Neural Networks for Intelligent Transportation Systems:

    2023

  53. [53]

    Physics-Informed Neural Networks:

    Raissi, Maziar and Perdikaris, Paris and Karniadakis, George E , year = 2019, journal =. Physics-Informed Neural Networks:

    2019

  54. [54]

    Advances in Neural Information Processing Systems , volume =

    Recipe for a General, Powerful, Scalable Graph Transformer , author =. Advances in Neural Information Processing Systems , volume =

  55. [55]

    Selfish Routing and the Price of Anarchy , author =

  56. [56]

    Distill , volume =

    A Gentle Introduction to Graph Neural Networks , author =. Distill , volume =

  57. [57]

    IEEE transactions on neural networks , volume =

    The Graph Neural Network Model , author =. IEEE transactions on neural networks , volume =

  58. [58]

    Sch. Schnet:. Advances in neural information processing systems , volume =

  59. [59]

    Urban Transportation Networks , author =

  60. [60]

    Transportation Networks for Research , author =

  61. [61]

    Bstabler/

    Stabler, Ben , year = 2026, month = mar, urldate =. Bstabler/

    2026

  62. [62]

    Advances in neural information processing systems , volume =

    Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domains , author =. Advances in neural information processing systems , volume =

  63. [63]

    Advances in neural information processing systems , volume =

    Attention Is All You Need , author =. Advances in neural information processing systems , volume =

  64. [64]

    arXiv preprint arXiv:1710.10903 , eprint =

    Graph Attention Networks , author =. arXiv preprint arXiv:1710.10903 , eprint =

    Pith/arXiv arXiv

  65. [65]

    Some Theoretical Aspects of Road Traffic Research

    Road Paper. Some Theoretical Aspects of Road Traffic Research. , author =. Proceedings of the institution of civil engineers , volume =

  66. [66]

    Foss4g North America , year =

    AequilibraE: a free QGIS add-on for transportation modeling , author =. Foss4g North America , year =

  67. [67]

    Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence,

    Graph WaveNet for Deep Spatial-Temporal Graph Modeling , author =. Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence,. 2019 , month =. doi:10.24963/ijcai.2019/264 , url =

    doi:10.24963/ijcai.2019/264 2019

  68. [68]

    Transportation Research Part C: Emerging Technologies , author =

    Use of graph attention networks for traffic assignment in a large number of network scenarios , volume =. Transportation Research Part C: Emerging Technologies , author =. 2025 , keywords =. doi:10.1016/j.trc.2025.104997 , abstract =

    arXiv 2025

  69. [69]

    Transportation Research Part C: Emerging Technologies , author =

    From optimization to prediction:. Transportation Research Part C: Emerging Technologies , author =. 2026 , keywords =. doi:10.1016/j.trc.2026.105808 , abstract =

    arXiv 2026