Pith. sign in

REVIEW 4 major objections 5 minor 46 references

Prediction of the Most Fire-Sensitive Point in Building Structures with Differentiable Agents for Thermal Simulators

T0 review · 4 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read A graph neural network trained on finite-element fire simulations can pin down the single fire location that would most damage a steel building, and it does so in milliseconds.

desk verdict Useful applied paper with a genuinely new problem formulation, but the headline MFSP accuracies are measured against the surrogate's own pseudo-labels; the single FEA check is encouraging but not enough. read the letter →

arxiv 2502.03424 v5 pith:LOVQE2NB submitted 2025-02-05 cs.LG

Yuan Xinjie , Khalid M. Mosalam This is my paper · ORCID
classification cs.LG
keywords MostFire-SensitivePointGraphNeuralNetworkMaximumInterstoryDriftRatioFiresafetyassessmentFiniteelementanalysissurrogateTransferlearningPseudo-labelingSteelframestructures
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper introduces the Most Fire-Sensitive Point (MFSP) of a building, defined as the fire origin that would cause the largest structural damage as measured by the Maximum Interstory Drift Ratio (MIDR). It claims that a two-network machine-learning pipeline can find this point without brute-force simulation of every possible fire: a graph neural network first learns to imitate finite element analysis (FEA) of fire scenarios, and a second network is trained as an 'argmaxer' over that surrogate to output the worst-case fire point directly. On a large synthetic dataset of steel frames the surrogate's MIDR ranking correlates with FEA at 0.74 on average and above 0.91 for severe fire cases, while the MFSP predictor reaches 74.1% room-level Top-5 accuracy with millisecond inference. If the claim holds, a fire-safety pre-check could simulate only the single worst-case fire scenario instead of enumerating all candidate locations.

What carries the argument

The machinery is a message-passing graph neural network in which beams and columns are edges and their joints are nodes, together with an edge-update module that refreshes each edge vector from its two endpoint node embeddings at every layer so that the graph itself degrades as the fire develops. Node features encode the extended coordinates of both the structural node and the fire source, their differences and Euclidean distance. For the MFSP task, a randomized virtual fire point occupies the fire-source input slot during training, forcing the network to rely on global structure; the frozen MIDR predictor generates pseudo ground-truth labels at room centers, and the hybrid loss combines the negative surrogate output with an MSE term on those pseudo-coordinates.

What would settle it

Run brute-force FEA over all candidate fire points in several held-out buildings (not just the 30 sampled scenarios), take the true MFSP as the FEA argmax, and compare both the MIDR surrogate's top point and the MFSP predictor's output against it; the claim is falsified if room-level Top-5 accuracy against true FEA labels is far below the 74.1% reported against pseudo-labels.

Watch

Extended reading notes

Core claim

The central claim is that the worst-case fire location for a framed steel building can be predicted as an optimization over a learned surrogate rather than by physical brute force. The MIDR predictor, a message-passing GNN whose edge attributes are updated each layer to reflect fire-induced degradation, is trained on FEA-generated labels and then frozen. The MFSP predictor reuses that GNN as a feature extractor and is trained, using pseudo-labels obtained by evaluating the surrogate at every room center plus a differentiable loss that maximises predicted MIDR, to output the fire point that maximises the surrogate's MIDR. The paper reports room-level Top-5 accuracy up to 74.1%, and argues that the transfer-learning initialization and the edge-update mechanism are each responsible for a measurable share of that performance.

Load-bearing premise

The load-bearing premise is that the MIDR surrogate's ranking of fire locations is faithful to the true finite-element ranking; the reported accuracies measure agreement with the surrogate's own pseudo-labels, and only one building is checked against brute-force FEA.

Editorial extensions

If this is right

  • Fire-safety pre-checks during preliminary design could run the worst-case fire scenario only, replacing exhaustive enumeration of fire origins with a single simulation.
  • For buildings where any fire is severe (some fire case exceeds 2% MIDR), the surrogate's ranking correlation exceeds 0.91, so the identified worst-case point is most reliable exactly in the cases that matter.
  • Because the surrogate computes MIDR for all scenarios in milliseconds per batch versus tens of seconds per scenario for FEA, iterative design exploration becomes interactive.
  • Since the framework is agnostic to the performance metric, replacing MIDR by another damage indicator extends the same pipeline to other failure modes.
  • The reported gains of edge-update and transfer learning imply that physics-relevant graph updates and pre-trained structural representations are worth keeping in later versions of the framework.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The reported Top-5 accuracy compares the MFSP predictor with the MIDR surrogate's pseudo-labels, not with FEA; the evaluation would be strengthened by measuring both networks against brute-force FEA on several buildings, as done once for ex2.
  • The differentiable-agent-plus-argmaxer recipe transfers to any problem of finding an input that maximizes an expensive simulator output, such as the most contaminant-sensitive point in a ventilation layout.
  • A direct test of the pseudo-labeling assumption is to compute the Spearman correlation between the surrogate's MIDR rankings and the FEA rankings on the existing labeled set; if it drops sharply for non-severe buildings, the practical usefulness of the pipeline is concentrated in severe cases.
  • Because the inference-time virtual fire point is the building's geometric center, one could probe robustness by testing buildings whose true MFSP lies at that center; if the predictor systematically errs there, the VFP trick is biasing the learned argmax.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper introduces the Most Fire-Sensitive Point (MFSP) concept—the fire location that maximizes the Maximum Interstory Drift Ratio (MIDR) of a building—and proposes a two-stage GNN framework to identify it. In stage 1, a GNN-based MIDR predictor is trained on FEA (χara) simulations; in stage 2, this predictor acts as a differentiable surrogate that pseudo-labels unlabeled structures and guides the training of an MFSP predictor through MSE and hybrid losses. The authors also propose a Transfer Learning scheme and an Edge Update mechanism. Experiments on a generated dataset of 1,573 labeled and 16,050 unlabeled structures report MIDR Spearman correlations up to 0.742 (over 0.91 for severe cases) and room-level top-5 MFSP accuracy up to 74.06%. The central claim is that the framework efficiently identifies the worst-case fire location, enabling fire safety pre-checks to simulate only that scenario.

Significance. If the reported accuracy were measured against true FEA ground truth, the paper would be a valuable contribution to fire safety assessment in preliminary structural design. The strengths of the manuscript include a clearly defined problem, an open-sourced dataset and code, a nontrivial GNN architecture with an edge-update mechanism, and a three-orders-of-magnitude speedup relative to FEA. The single external FEA case study (ex2) showing a 0.8 m error for the hybrid loss is encouraging. However, the headline MFSP accuracies are computed against pseudo-labels generated by the same MIDR surrogate that also supervises the MFSP predictor, so the central claim lacks population-level external validation. The need for such validation is in fact acknowledged by the authors in Section VIII-B, which states that rigorous experimental validation and cross-comparison studies are essential.

major comments (4)
  1. [Section VII-B, Eqs. (11), (23), (24), Table VII] The MFSP predictor is trained and evaluated entirely against pseudo ground truths produced by the MIDR predictor. Equation (11) trains the MFSP network on pseudo-labeled MFSP coordinates, and Eqs. (23)-(24) and Table VII compute distance error, room distance error, and room rank against those same pseudo-labels. Consequently, the reported top-5 accuracy of 74.06% and the average room rank of about 6.7 measure consistency between the MFSP network and the MIDR surrogate, not agreement with FEA. This is a teacher-student loop that does not establish the central claim. Please add an external evaluation on a random sample of test structures in which FEA is used to compute the true MFSP (e.g., by simulating all 30 room-center fire scenarios, or a grid of points), and report distance error, room rank, and top-5 accuracy against those FEA ground truths. Also report the agreement between the MIDR predictor's argmax and the FEA argmax, especially for structures where several rooms have similar MIDR values.
  2. [Section VII-A3, Table V, Fig. 9] An average Spearman correlation of 0.74, or even 0.91 for severe cases, is not by itself sufficient to certify argmax fidelity. Spearman correlation is computed over 30 fire scenarios per structure and does not quantify how often the surrogate's highest-MIDR point coincides with the FEA highest-MIDR point; a correlation of 0.91 can still misorder the top several positions, and the MFSP is defined by the argmax. Please report the exact room-level MFSP match rate of the MIDR predictor against FEA, and characterize the distribution of MIDR gaps between the top-ranked and second-ranked fire points. Without such information, the statement in Section VII-A3 that 'the high value of average ρs ≥ 0.90 guarantees the reliability and validity' is not supported.
  3. [Section VII-B2, case study ex2] The only brute-force FEA validation of the MFSP predictor is the single building ex2, with 1,232 grid points and one reported hybrid-loss error of 0.8 m. A single example cannot establish population-level accuracy, and the text does not report replication or sensitivity to the 2 m grid spacing. Please extend the FEA-based validation to a sample of structures, preferably including severe cases across the test set, and report both MSE-only and hybrid losses for all of them. This would also allow a direct comparison of the two loss functions against true FEA ground truth rather than against pseudo-labels.
  4. [Section VI-D and Section VIII-B] The rule-based thermal load generation model, with hand-chosen parameters r_up, r_down, α_i, and β_i, is an ad hoc approximation of fire propagation. Because the FEA labels themselves are generated from this model, the entire pipeline is validated only against this simplified physics. The manuscript acknowledges in Section VIII-B that 'rigorous experimental validation and cross-comparison studies are essential to verify simulation fidelity,' which is an important limitation. The abstract and conclusions should therefore scope the central claim to structures and thermal fields of the type generated by the authors' rule-based model, and the paper should state explicitly that real-world fire validation is absent.
minor comments (5)
  1. [Section III-B] The word 'identifing' should be 'identifying' in the description of the MFSP predictor.
  2. [Table VIII and surrounding text] 'Hybird' appears instead of 'Hybrid' in the table caption and in the discussion of CDF values.
  3. [Section IV and Section VII] The acronym 'MSFP predictor' appears in a few places where 'MFSP predictor' is meant; please correct the transposed spelling.
  4. [Section VII-A1] The word 'implemente' should be 'implemented' in the sentence describing the two versions of the MIDR predictor.
  5. [Figure 9 caption and Section VII-A3] The caption labels panel (a) as 'CCDF' while the text refers to a CDF; please standardize the terminology.

Circularity Check

1 steps flagged · score 6.0 of 10

The headline MFSP accuracies are computed against pseudo ground truths generated by the same MIDR surrogate used to train the MFSP predictor; only one FEA check (ex2) breaks the loop.

  1. fitted input called prediction [Section V-A, Eqs. (10)-(12); Section VII-B1, Eq. (23) and room-rank definition]
    "Instead of directly using L_MIDR, we use the MIDR predictor to generate pseudo ground truths for the MFSP. ... the one with the highest MIDR is selected as the pseudo ground truth MFSP for the building at hand. ... These pseudo-labels are used to evaluate the MFSP predictor through various metrics ... generated by selecting the fire point with the highest predicted MIDR for each structure."

    The pseudo ground truth MFSP is, by the paper's own construction, the argmax of the MIDR surrogate over room centers. Training the MFSP predictor with Eq. (11) and scoring it with Eq. (23) against those same pseudo labels measures how closely the MFSP network reproduces the surrogate's argmax, not how well it finds the FEA-defined MFSP. The reported top-5 accuracy (74.1%), room rank, and distance errors are therefore teacher-student consistency numbers: the evaluation target is generated by the same model family whose behavior the MFSP predictor is trained to imitate. The only FEA-based MFSP verification is the single ex2 building, so the population-level claim is not independently anchored.

full rationale

The chain is not wholly self-referential: the MIDR predictor is trained on 47,190 FEA fire cases from χara, and the ex2 brute-force FEA check (1,232 grid points) provides one external anchor. However, the MFSP predictor's headline evaluation is entirely against pseudo ground truths produced by that same MIDR surrogate (Eq. 23 and room-rank definition in Section VII-B1), so the reported accuracies measure agreement with the surrogate's argmax rather than with FEA. This is a partial circularity: the 'prediction' being scored is the surrogate's argmax except in the single ex2 case. No load-bearing self-citation chain, imported uniqueness theorem, or ansatz-smuggling was found; the self-citations to χara and prior STAIRlab tools are ordinary tool usage, not circular evidence. The correct response to this circularity is to re-run the MFSP evaluation on brute-force FEA MFSP labels for a sample of structures, or at least to threshold on surrogate confidence.

Assumptions & free parameters 6 free parameters · 6 assumptions · 1 invented entities

The central workflow rests on several hand-chosen simulation parameters (fire spread coefficients, loss weights, filtering thresholds) and on strong domain simplifications (uniform thermal loads, small deformations, MIDR as stability metric). No new physical entities are postulated; the VFP is a computational artifact. The pseudo-labeling loop is the most consequential methodological choice, as it makes the MFSP evaluation self-referential.

free parameters (6)
  • r_up = 0.95
    Ratio controlling upward fire spread in the t1 transition-time formula (Eq. 18, Table III). Chosen by hand to mimic expected fire spread; no calibration data or sensitivity analysis.
  • r_down = 0.97
    Ratio controlling downward fire spread in Eq. 18, Table III. Chosen by hand, same rationale as r_up.
  • alpha_1, alpha_2, alpha_3 = 10, 18, 5
    Length-scale constants in the t1 transition-time relationships (Eq. 18). Fixed constants that determine how distance affects fire arrival time; not derived from experiments.
  • beta_1, beta_2, beta_3 = 16, 18, 30
    Time-scale constants in Eq. 18 and Table III; beta_1 and beta_3 are scaled by geometric series in story difference. Chosen to reproduce qualitative fire spread behavior, not fitted to data.
  • w1 (hybrid loss weight) = 10, 50, 100 (w2=1)
    Weight balancing the MIDR loss and MSE loss in Eq. 12. Varied in experiments; the final reference uses w1=50. These are hyperparameters, not model parameters.
  • MIDR filter threshold = 1%
    Structures with MIDR above 1% under gravity loads are excluded from the dataset. A design choice that affects the population and could influence the difficulty of the learned task.
assumptions (6)
  • domain assumption Steel structures rapidly equilibrate with surrounding gas temperatures, so gas temperature can be used directly as uniform element temperature in FEA.
    Invoked in Section VI-D as the basis for the rule-based thermal load generation. Cited literature supports this for thin steel sections, but it is an approximation.
  • domain assumption ISO 834 standard fire curve represents the burning stage of a fire for all elements.
    Used in Eq. 15 for the second temperature stage. Standard in fire engineering, but real fires vary with ventilation and fuel load; the paper acknowledges this simplification.
  • domain assumption Temperature-dependent material reduction factors from EN 1993-1-2 apply.
    Implemented in Steel01Thermal material model in Section VI-E. This is standard, but the paper does not validate it against experimental fire tests.
  • domain assumption Small displacements and rotations, with linear geometric transformations, are sufficient for fire response.
    Explicitly stated in Section VI-E: 'we did not consider large deformations'. This excludes buckling and P-delta effects, which can be critical in fire.
  • domain assumption MIDR is an adequate proxy for overall fire-induced structural instability.
    Section III-A justifies MIDR as a proxy while acknowledging it misses local buckling, midspan softening, and connection failures. The claim that MFSP identifies a 'most fire-sensitive point' depends on this metric being the right one.
  • ad hoc to paper The rule-based thermal load generation with hand-chosen parameters approximates real fire propagation.
    Section VI-D describes the parameter choice as 'inspired by' fire dynamics and cites qualitative agreement with a Markov-chain model. No calibration to experimental data or sensitivity analysis is provided.
invented entities (1)
  • Virtual Fire Point (VFP)
    purpose: A placeholder fire location used as GNN input during MFSP predictor training, masking fire-specific features so the network learns global structural information.
    Introduced in Section V-B. It is a training trick with no empirical handle. During inference, VFP is set to the geometric center of the building.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Prediction of the Most Fire-Sensitive Point in Building Structures with Differentiable Agents for Thermal Simulators." pith.science (2026). https://pith.science/paper/LOVQE2NB

@misc{pith2026250203424,
  author       = {Pith},
  title        = {Pith review of: Prediction of the Most Fire-Sensitive Point in Building Structures with Differentiable Agents for Thermal Simulators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LOVQE2NB}},
  note         = {Machine review of arXiv:2502.03424}
}
read the original abstract

Fire safety is crucial for ensuring the stability of building structures, yet evaluating whether a structure meets fire safety requirement is challenging. Fires can originate at any point within a structure, and simulating every potential fire scenario is both expensive and time-consuming. To address this challenge, we propose the concept of the Most Fire-Sensitive Point (MFSP) and an efficient machine learning framework for its identification. The MFSP is defined as the location at which a fire, if initiated, would cause the most severe detrimental impact on the building's stability, effectively representing the worst-case fire scenario. In our framework, a Graph Neural Network (GNN) serves as an efficient and differentiable agent for conventional Finite Element Analysis (FEA) simulators by predicting the Maximum Interstory Drift Ratio (MIDR) under fire, which then guides the training and evaluation of the MFSP predictor. Additionally, we enhance our framework with a novel edge update mechanism and a transfer learning-based training scheme. Evaluations on a large-scale simulation dataset demonstrate the good performance of the proposed framework in identifying the MFSP, offering a transformative tool for optimizing fire safety assessments in structural design. All developed datasets and codes are open-sourced online.

Figures

Figures reproduced from arXiv: 2502.03424 by the authors.

Figure 1
Figure 1. Proposed framework for predicting the MFSP in building structures. Trapezoids and gray rectangles represent the NN [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Illustration of a K-layer GNN architecture, demonstrating information flow within the network for a graph with N nodes. attributes. Additionally, its floor level hi , a critical factor to define the fire scenarios, is included as another node attribute. For ease of description, we call the four-tuple (xi , yi , zi , hi) extended coordinates of node i. To account for the fire loca￾tion, the extended coordinates of th… view at source ↗
Figure 3
Figure 3. Different training strategies (head-1 & head-2 are MLPs [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Procedure of pseudo-labeling for building structure [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Workflow for dataset generation (geometry, material property, gravity loads, and fire scenarios). [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Examples of generated structural geometry of different sizes (all dimensions in meters). [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: illustrates the t1 curves (t1 in minutes vs. Ls or Lp in meters) for various fire scenarios: (1) fire originating on the lower floor, h − hf = 1 with rapid upward spread, (2) fire on the same floor, h = hf with the fastest spread, and (3) fire on the upper floor, hf − …
Figure 9
Figure 9. Figure 9: Representative MIDR predictor results: (a) CDF of [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]
Figure 11
Figure 11. Figure 11: Illustrations of positions of ground truth and predicted MFSPs in three views of the building of [PITH_FULL_IMAGE:figures/full_fig_p021_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 46 canonical work pages

  1. [1]

    Rokibul Alam, Nazmul Siddique, and Hojjat Adeli

    Kazi Md. Rokibul Alam, Nazmul Siddique, and Hojjat Adeli. A dynamic ensemble learning algorithm for neural networks.Neural Computing and Applications, 32(12):8675–8690, June 2020

    work page 2020

  2. [2]

    ASCE, Reston, V A, 2016

    American Society of Civil Engineers.Minimum Design Loads and Associated Criteria for Buildings and Other Structures (ASCE/SEI 7- 16). ASCE, Reston, V A, 2016

    work page 2016

  3. [3]

    Modelling steel behaviour.Fire Safety Journal, 13(1):17–26, 1988

    Yngve Anderberg. Modelling steel behaviour.Fire Safety Journal, 13(1):17–26, 1988

    work page 1988

  4. [4]

    Srinivasan Arunachalam and Seymour M. J. Spence. Generalized strat- ified sampling for efficient reliability assessment of structures against natural hazards.Journal of Engineering Mechanics, 149(7):04023042, 2023

    work page 2023

  5. [5]

    Statis- tics on CA wildfires and CAL FIRE activity

    California Department of Forestry and Fire Protection, Cal Fire. Statis- tics on CA wildfires and CAL FIRE activity. https://www.fire.ca.gov/ stats-events/, Accessed Jan. 9, 2025, 2025

    work page 2025

  6. [6]

    Huaguo Chen, Jianjun Yang, Xinhong Chen, Dong Zhang, and Vincent J. L. Gan. Tempnet: A graph convolutional network for temperature field prediction of fire-damaged concrete.Expert Systems with Applications, 238:121997, March 2024

    work page 2024

  7. [7]

    Hadjisophocleous

    Hao Cheng and George V . Hadjisophocleous. Dynamic modeling of fire spread in building.Fire Safety Journal, 46(4):211–224, May 2011

    work page 2011

  8. [8]

    Jean, Kai-Hung Chang, Yin-Nan Huang, and Chuin-Shan Chen

    Yuan-Tung Chou, Wei-Tze Chang, Jimmy G. Jean, Kai-Hung Chang, Yin-Nan Huang, and Chuin-Shan Chen. StructGNN: An efficient graph neural network framework for static structural.Computers & Structures, 299:107385, August 2024

    work page 2024

Show all 46 references
  1. [9]

    A new temperature–time curve for fire- resistance analysis of structures.Fire Safety Journal, 54:113–120, November 2012

    Yong Du and Guo-qiang Li. A new temperature–time curve for fire- resistance analysis of structures.Fire Safety Journal, 54:113–120, November 2012

    2012

  2. [10]

    Fire load and fire growth character- istics in modern high-rise buildings.Fire Safety Journal, 135:103710, February 2023

    Ugur Dundar and Serdar Selamet. Fire load and fire growth character- istics in modern high-rise buildings.Fire Safety Journal, 135:103710, February 2023

    2023

  3. [11]

    EN 1993-1-2:2005 (E), 2005

    European Committee for Standardization. EN 1993-1-2:2005 (E), 2005. Available from the European Committee for Standardization (CEN)

    1993

  4. [12]

    Thermal examples, 2022

    Erica Fischer and Walker Maddalozzo. Thermal examples, 2022

    2022

  5. [13]

    Fire resistance assessment of concrete structures.Fire Safety Journal, 37(3):241–273, 2002

    Jean-Marc Franssen and Venkatesh K Kodur. Fire resistance assessment of concrete structures.Fire Safety Journal, 37(3):241–273, 2002

    2002

  6. [14]

    John Wiley & Sons, Ltd, 2016

    Jean-Marc Franssen and Paulo Vila Real.Fire Design of Steel Structures: EC1: Actions on structures; Part 1-2: Actions on structure exposed to fire; EC3: Design of steel structures; Part 1-2: Structural fire design, 2nd Edition. John Wiley & Sons, Ltd, 2016

    2016

  7. [15]

    Huanxiang Gao, Gang Hu, Dongqin Zhang, Wenjun Jiang, K. T. Tse, K. C. S. Kwok, and Ahsan Kareem. Urban wind field prediction based on sparse sensors and physics-informed graph-assisted auto-encoder. Computer-Aided Civil and Infrastructure Engineering, 39(10):1409– 1430, 2024

    2024

  8. [16]

    Multi-agent replicator controller for sustainable vibration control of smart structures.Journal of Vibroengineering, 19(6):4300–4322, September 2017

    Mariantonieta Gutierrez Soto and Hojjat Adeli. Multi-agent replicator controller for sustainable vibration control of smart structures.Journal of Vibroengineering, 19(6):4300–4322, September 2017

    2017

  9. [17]

    Inductive representation learning on large graphs.NeurIPS, 2017

    Will L Hamilton, Rex Ying, and Jure Leskovec. Inductive representation learning on large graphs.NeurIPS, 2017

    2017

  10. [18]

    Study on fire plume and surface temperature field of components in mid through steel box ribbed arch bridge under tanker fire.Fire Technology, December 2024

    Zhongying He, Jianxiang Man, Guolin Li, Haoyang Li, Yifan Song, Tianxing Liu, and Chaojie Song. Study on fire plume and surface temperature field of components in mid through steel box ribbed arch bridge under tanker fire.Fire Technology, December 2024

    2024

  11. [19]

    Hokugo, Y

    A. Hokugo, Y . Hasemi, Y . Hayashi, and Masashi Yoshida. Mechanism for the upward fire spread through balconies based on an investigation and experiments for a multi-story fire in high-riseapartment building. Fire Safety Science, 6:649–660, 2000

    2000

  12. [20]

    A dynamic graph deep learning model with multivariate empirical mode decomposition for network-wide metro passenger flow prediction

    Hao Huang, Jiannan Mao, Leilei Kang, Weike Lu, Sijia Zhang, and Lan Liu. A dynamic graph deep learning model with multivariate empirical mode decomposition for network-wide metro passenger flow prediction. Computer-Aided Civil and Infrastructure Engineering, 39(17):2596– 2618, 2024

    2024

  13. [21]

    ISO, Geneva, Switzerland, 1999

    International Organization for Standardization.Fire resistance tests – Elements of building construction – Part 1: General requirements. ISO, Geneva, Switzerland, 1999. ISO 834-1:1999

    1999

  14. [22]

    Reza Hosseini, Wenchi Shou, Peng Wu, and Chao Mao

    Yilong Jia, Jun Wang, M. Reza Hosseini, Wenchi Shou, Peng Wu, and Chao Mao. Temporal graph attention network for building thermal load prediction.Energy and Buildings, 321:113507, October 2024

    2024

  15. [23]

    Reza Hosseini, and Yu Bai

    Yilong Jia, Jun Wang, Wenchi Shou, M. Reza Hosseini, and Yu Bai. Graph neural networks for construction applications.Automation in Construction, 154:104984, October 2023

    2023

  16. [24]

    OpenSees software architecture for the analysis of structures in fire.Journal of Computing in Civil Engineering, 29(1):04014030, 2015

    Jian Jiang, Liming Jiang, Panagiotis Kotsovinos, Jian Zhang, Asif Usmani, Frank McKenna, and Guo-Qiang Li. OpenSees software architecture for the analysis of structures in fire.Journal of Computing in Civil Engineering, 29(1):04014030, 2015

    2015

  17. [25]

    Response of steel beam–columns exposed to fire.Engineering Structures, 32(2):607–619, 2010

    Venkatesh K Kodur and Mazen MB Dwaikat. Response of steel beam–columns exposed to fire.Engineering Structures, 32(2):607–619, 2010

    2010

  18. [26]

    Lee and K

    T.-H. Lee and K. M. Mosalam. Probabilistic fiber element modeling of reinforced concrete structures.Computers and Structures, 82(27):2285– 2299, 2004

    2004

  19. [27]

    FDS results for selecting the right scenario in the case of a building fire: A case study

    Florin Manea, Emilian Ghicioi, Marius Cornel Suvar, Maria Prodan, Nicolae Ioan Vlasin, Niculina Sonia Suvar, and Titus Vlase. FDS results for selecting the right scenario in the case of a building fire: A case study. Fire, 5(6), 2022

    2022

  20. [28]

    Fire dynamics simulator- technical reference guide

    Kevin B McGrattan, Howard R Baum, Ronald G Rehm, Anthony Hamins, and Glenn P Forney. Fire dynamics simulator- technical reference guide. Technical Report NIST IR 6467, National Institute of Standards and Technology, Gaithersburg, MD, 2000

    2000

  21. [29]

    Fenves, and Michael H

    Frank McKenna, Gregory L. Fenves, and Michael H. Scott. OpenSees: A framework for earthquake engineering simulation.Computing in Science & Engineering, 13(4):58–66, 2011

    2011

  22. [30]

    S. A. Mirza and J. G. MacGregor. Variations in dimensions of reinforced concrete members.Journal of Structural Division, ASCE, 105(ST4):751–766, 1979

    1979

  23. [31]

    Structural-fire responses forecasting via modular AI.Fire Safety Journal, 140:103863, October 2023

    Zhuojun Nan, Mhd Anwar Orabi, Xinyan Huang, Yaqiang Jiang, and Asif Usmani. Structural-fire responses forecasting via modular AI.Fire Safety Journal, 140:103863, October 2023

    2023

  24. [32]

    Perez, F

    C. Perez, F. McKenna, and K. M. Mosalam. OpenSeesRT, version 0.0.63, 2024

    2024

  25. [33]

    Claudio Perez, Frank McKenna, and Khalid M. Mosalam. χara, July 2024

    2024

  26. [34]

    Review of design approaches for steel columns subjected to localised fires: Insights from the 2nd generation eurocode model.Engineering Structures, 315:118424, 2020

    Luca Possidente, Carlos Couto, Nicola Tondini, and Paulo Vila Real. Review of design approaches for steel columns subjected to localised fires: Insights from the 2nd generation eurocode model.Engineering Structures, 315:118424, 2020

    2020

  27. [35]

    Rafiei, Lynne V

    Mohammad H. Rafiei, Lynne V . Gauthier, Hojjat Adeli, and Daniel Takabi. Self-Supervised Learning for Electroencephalography.IEEE Transactions on Neural Networks and Learning Systems, 35(2):1457– 1471, February 2024

    2024

  28. [36]

    Rafiei, Lynne V

    Mohammad H. Rafiei, Lynne V . Gauthier, Hojjat Adeli, and Daniel Takabi. Self-Supervised Learning for Near-Wild Cognitive Workload Estimation.Journal of Medical Systems, 48(1):107, November 2024

    2024

  29. [37]

    Ego-planning-guided multi-graph convolutional network for heterogeneous agent trajectory prediction.Computer-Aided Civil and Infrastructure Engineering, 39(22):3357–3374, 2024

    Zihao Sheng, Zilin Huang, and Sikai Chen. Ego-planning-guided multi-graph convolutional network for heterogeneous agent trajectory prediction.Computer-Aided Civil and Infrastructure Engineering, 39(22):3357–3374, 2024

    2024

  30. [38]

    Centre for Advanced Engineering, University of Canterbury, 2008

    Michael Spearpoint.Fire Engineering Design Guide. Centre for Advanced Engineering, University of Canterbury, 2008

    2008

  31. [39]

    MIT Press, 2018

    Richard Sutton and Andrew Barto.Reinforcement Learning: An Intro- duction. MIT Press, 2018

    2018

  32. [40]

    Graph attention networks.ICLR, 2018

    Petar Veli ˇckovi´c, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Li `o, and Yoshua Bengio. Graph attention networks.ICLR, 2018

    2018

  33. [41]

    Tongtong Wan, Hainian Wang, Xu Yang, Yu Chen, Lian Li, and Aboelkasim Diab. Thermal contraction coordination behavior between unbound aggregate layer and asphalt mixture overlay based on the finite difference and discrete element coupling method.Computer-Aided Civil and Infras...

    2024

  34. [42]

    Optimization-improved thermal– mechanical simulation of welding residual stresses in welded con- nections.Computer-Aided Civil and Infrastructure Engineering, 39(9):1275–1293, 2024

    Le Wang and Xudong Qian. Optimization-improved thermal– mechanical simulation of welding residual stresses in welded con- nections.Computer-Aided Civil and Infrastructure Engineering, 39(9):1275–1293, 2024

    2024

  35. [43]

    Yao Wang, Guo-Qiang Li, and Shaojun Zhu. Graph recurrent neural networks-integrated real-time prediction of key displacements for fire- induced collapse early warning of steel frames.Applied Soft Computing, 163:111942, September 2024

    2024

  36. [44]

    L. Yao, Z. Leng, J. Jiang, and F. Ni. A multi-agent reinforcement learning model for maintenance optimization of interdependent highway pavement networks.Computer-Aided Civil and Infrastructure Engineer- ing, 39(19):2951–2970, 2024

    2024

  37. [45]

    Yu and H

    G. Yu and H. Adeli. Object-Oriented Finite Element Analysis Using EER Model.Journal of Structural Engineering, 119(9):2763–2781, September 1993

    1993

  38. [46]

    Guowei Zhang, Huaxiang Li, Guoqing Zhu, and Junyi Li. Temperature fields for fire resistance analysis of structures exposed to natural fires in large space buildings.The Structural Design of Tall and Special Buildings, 29(4):e1708, 2020

    2020

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.