pith. sign in

arxiv: 2511.03015 · v2 · submitted 2025-11-04 · 💻 cs.LG · stat.ML

Discrete Bayesian Sample Inference for Graph Generation

Ole Petersen , Marcel Kollovieh , Marten Lienen , Stephan G\"unnemann This is my paper

Pith reviewed 2026-05-18 00:40 UTC · model grok-4.3

classification 💻 cs.LG stat.ML
keywords graph generationbayesian sample inferencediscrete diffusionmolecular graphsone-shot generationstochastic differential equationsscore approximation
0
0 comments X p. Extension

The pith

GraphBSI generates discrete graphs by refining beliefs over distribution parameters in continuous space.

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

The paper introduces GraphBSI, a one-shot generative model for graphs that works by iteratively refining a belief over possible graphs instead of changing graph samples directly. This belief is represented in the continuous space of distribution parameters, which allows the method to manage the discrete and unordered character of graph data more naturally. The authors express Bayesian Sample Inference as a stochastic differential equation and derive a controlled-noise family of such equations that maintains the required marginal distributions through an approximation to the score function. Their analysis links the approach to Bayesian Flow Networks and diffusion models. Empirical tests show the method reaching state-of-the-art results on molecular and synthetic graph benchmarks.

Core claim

GraphBSI is a one-shot graph generative model based on Bayesian Sample Inference. Instead of evolving samples directly, it iteratively refines a belief over graphs in the continuous space of distribution parameters. BSI is stated as a stochastic differential equation, and a noise-controlled family of SDEs is derived that preserves the marginal distributions via an approximation of the score function. This formulation reveals connections to Bayesian Flow Networks and diffusion models while delivering state-of-the-art performance on molecular and synthetic graph generation.

What carries the argument

Bayesian Sample Inference (BSI), which iteratively refines a belief over graphs in the continuous space of distribution parameters rather than evolving the discrete graph samples themselves.

If this is right

  • It outperforms existing one-shot graph generative models on the Moses and GuacaMol benchmarks for molecular and synthetic graphs.
  • It handles the discrete and unordered nature of graphs by operating on continuous parameter beliefs.
  • It establishes a theoretical link between Bayesian Sample Inference, Bayesian Flow Networks, and diffusion models.
  • The derived noise-controlled SDEs maintain the marginal distributions needed for consistent generation.

Where Pith is reading between the lines

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

  • The continuous-belief approach could extend to generating other discrete objects such as sequences or point sets.
  • Initial belief parameters could be set to enforce desired global properties during generation.
  • The SDE perspective may suggest new hybrid sampling schemes that blend continuous refinement with discrete updates.

Load-bearing premise

The approximation of the score function used to derive the noise-controlled family of SDEs is accurate enough to preserve the required marginal distributions and support high-quality one-shot graph generation.

What would settle it

Training GraphBSI on the Moses dataset and then sampling many graphs whose property distributions, such as atom counts or ring sizes, deviate markedly from the training set statistics.

Figures

Figures reproduced from arXiv: 2511.03015 by Marcel Kollovieh, Marten Lienen, Ole Petersen, Stephan G\"unnemann.

Figure 1
Figure 1. Figure 1: Illustration of GraphBSI’s generative process. Nodes and edges are modeled as indepen [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Trajectories of the SDE Theorem 4 for different values of γ with three classes and fixed reconstruction fθ(zt, t) = ˆe2. At γ = 0, the sampler resembles a probability flow ODE as in flow matching. Increasing γ leads to noisier trajectories. At γ = 1, the original SDE in Theorem 3 is recovered, and increasing the noise further makes the trajectories even more volatile. The density function of the marginal d… view at source ↗
Figure 3
Figure 3. Figure 3: Normalized metrics (zero mean, unit variance) vs. noise level [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Performance change for changes in the non-uniform timestepping parameter [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Results on the QM9 dataset. Theorem 6. Fixing the prediction xˆ = fθ(zt, t) and the values β = β(t+∆t/2), β ′ = β ′ (t+∆t/2) in Eq. (7) in a time interval [t, t+∆t] yields an Ornstein-Uhlenbeck process with the exact marginal zt+∆t ∼ m + (zt − m)e −κ∆t + r γβ′ 2κ (1 − e−2κ∆t) · N (0, I), (67) where κ = (γ−1)β ′ 2(β0+β) , m = µ0 + (β + β ′/κ)xˆ. Proof. The SDE in Eq. (7) with fixed parameters β, β′ , xˆ is … view at source ↗
read the original abstract

Generating graph-structured data is crucial in applications such as molecular generation, knowledge graphs, and network analysis. However, their discrete, unordered nature makes them difficult for traditional generative models, leading to the rise of discrete diffusion and flow matching models. In this work, we introduce GraphBSI, a novel one-shot graph generative model based on Bayesian Sample Inference (BSI). Instead of evolving samples directly, GraphBSI iteratively refines a belief over graphs in the continuous space of distribution parameters, naturally handling discrete structures. Further, we state BSI as a stochastic differential equation (SDE) and derive a noise-controlled family of SDEs that preserves the marginal distributions via an approximation of the score function. Our theoretical analysis further reveals the connection to Bayesian Flow Networks and Diffusion models. Finally, in our empirical evaluation, we demonstrate state-of-the-art performance on molecular and synthetic graph generation, outperforming existing one-shot graph generative models on the standard benchmarks Moses and GuacaMol.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The manuscript introduces GraphBSI, a one-shot graph generative model based on Bayesian Sample Inference (BSI). It refines beliefs over graphs in the continuous space of distribution parameters to handle discrete unordered structures. The authors formulate BSI as an SDE and derive a noise-controlled family of SDEs that preserve marginal distributions via an approximation of the score function. They establish theoretical connections to Bayesian Flow Networks and diffusion models, and report state-of-the-art empirical performance on molecular and synthetic graph generation using the Moses and GuacaMol benchmarks.

Significance. If the SDE derivation and score approximation are valid, the framework could provide a principled way to unify Bayesian inference with continuous generative processes for discrete data, potentially strengthening connections between BSI, Bayesian Flow Networks, and diffusion models. The reported SOTA results on standard benchmarks indicate practical promise for molecular generation tasks, though this depends on the approximation preserving marginals without distortion.

major comments (2)
  1. [Theoretical analysis] Theoretical analysis section: The central claim that the noise-controlled family of SDEs preserves the required marginal distributions over graphs rests on an approximation of the score function when transitioning from the Bayesian Sample Inference process to the continuous formulation. No explicit error bounds or validation specific to discrete graph marginals are supplied, which is load-bearing for the one-shot generation guarantee and the claimed connections to existing models.
  2. [Empirical evaluation] Empirical evaluation section: The SOTA performance on Moses and GuacaMol is presented as outperforming existing one-shot models, but the manuscript supplies no derivations, experimental controls, or error analysis in the abstract and limited description; this makes it impossible to verify whether benchmark scores arise from the SDE construction itself rather than post-hoc tuning.
minor comments (2)
  1. [Introduction] The abstract and introduction could more clearly distinguish the novel contributions of GraphBSI from prior work on discrete diffusion and flow matching models, including specific self-citation details for the BSI formulation.
  2. [Method] Notation for the continuous space of distribution parameters and the mapping from discrete graphs should be defined more explicitly to aid readability for readers unfamiliar with BSI.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive feedback. We address each major comment below and indicate the changes planned for the revised manuscript.

read point-by-point responses

  1. Referee: [Theoretical analysis] Theoretical analysis section: The central claim that the noise-controlled family of SDEs preserves the required marginal distributions over graphs rests on an approximation of the score function when transitioning from the Bayesian Sample Inference process to the continuous formulation. No explicit error bounds or validation specific to discrete graph marginals are supplied, which is load-bearing for the one-shot generation guarantee and the claimed connections to existing models.

    Authors: We agree that the score-function approximation is central to the SDE derivation and that explicit error bounds would strengthen the one-shot guarantee. In the revision we will add a dedicated subsection deriving an L2 error bound on the marginal preservation under standard Lipschitz assumptions on the score, together with a discrete-graph-specific validation experiment that measures the total-variation distance between the approximated and exact marginals on small synthetic graphs. This will also make the links to Bayesian Flow Networks and diffusion models more precise. revision: yes

  2. Referee: [Empirical evaluation] Empirical evaluation section: The SOTA performance on Moses and GuacaMol is presented as outperforming existing one-shot models, but the manuscript supplies no derivations, experimental controls, or error analysis in the abstract and limited description; this makes it impossible to verify whether benchmark scores arise from the SDE construction itself rather than post-hoc tuning.

    Authors: The full experimental section already contains ablation studies that isolate the contribution of the noise-controlled SDE, multiple random seeds with reported standard deviations, and direct comparisons against the same one-shot baselines. To address the concern about verifiability, we will expand the main-text description of the experimental protocol, add an explicit control that disables the SDE noise schedule while keeping all other hyperparameters fixed, and include a short derivation showing how the reported metrics are computed from the model outputs. These additions will appear in both the main paper and the appendix. revision: partial

Circularity Check

0 steps flagged

Derivation chain self-contained; approximation of score function stated explicitly without reduction to inputs

full rationale

The paper introduces GraphBSI from Bayesian Sample Inference, states BSI as an SDE, and derives a noise-controlled family of SDEs that preserves marginal distributions via an approximation of the score function. This approximation is presented as a methodological step rather than a tautology. Theoretical connections to Bayesian Flow Networks and diffusion models are derived from the construction, and empirical results on Moses/GuacaMol are reported separately. No equations or self-citations in the provided text reduce a central claim to a fitted input, self-definition, or load-bearing prior work by the same authors. The derivation remains independent of the target results and does not rename known patterns or smuggle ansatzes via citation. This is the common case of an honest non-finding.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Review is limited to the abstract; the ledger therefore records only the high-level modeling choice visible in the text.

axioms (1)
  • domain assumption Approximation of the score function preserves marginal distributions for the derived family of SDEs.
    Invoked when the authors derive the noise-controlled SDEs that preserve marginals.
invented entities (1)
  • GraphBSI no independent evidence
    purpose: One-shot graph generative model based on Bayesian Sample Inference
    New model name and architecture introduced to handle discrete graphs via continuous belief refinement.

pith-pipeline@v0.9.0 · 5699 in / 1280 out tokens · 49111 ms · 2026-05-18T00:40:26.513426+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages · 8 internal anchors

  1. [1]

    Barrett, Scott Cameron, Bora Guloglu, Matthew Greenig, Charlie B

    Timothy Atkinson, Thomas D. Barrett, Scott Cameron, Bora Guloglu, Matthew Greenig, Charlie B. Tan, Louis Robinson, Alex Graves, Liviu Copoiu, and Alexandre Laterre. Protein sequence modelling with bayesian flow networks. Nature Communications, 16 0 (1): 0 3197, 2025. ISSN 2041-1723. doi:10.1038/s41467-025-58250-2. URL https://doi.org/10.1038/s41467-025-58250-2

    work page doi:10.1038/s41467-025-58250-2 2025

  2. [2]

    Johnson, Jonathan Ho, Daniel Tarlow, and Rianne van den Berg

    Jacob Austin, Daniel D. Johnson, Jonathan Ho, Daniel Tarlow, and Rianne van den Berg. Structured denoising diffusion models in discrete state-spaces, 2023. URL https://arxiv.org/abs/2107.03006

    work page arXiv 2023

  3. [3]

    Efficient and scalable graph generation through iterative local expansion, 2024

    Andreas Bergmeister, Karolis Martinkus, Nathanaël Perraudin, and Roger Wattenhofer. Efficient and scalable graph generation through iterative local expansion, 2024. URL https://arxiv.org/abs/2312.11529

    work page arXiv 2024

  4. [4]

    Segler, and Alain C

    Nathan Brown, Marco Fiscato, Marwin H.S. Segler, and Alain C. Vaucher. Guacamol: Benchmarking models for de novo molecular design. Journal of Chemical Information and Modeling, 59 0 (3): 0 1096–1108, March 2019. ISSN 1549-960X. doi:10.1021/acs.jcim.8b00839. URL http://dx.doi.org/10.1021/acs.jcim.8b00839

    work page doi:10.1021/acs.jcim.8b00839 2019

  5. [5]

    Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, … Ilya Sutskever, et al

    Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, … Ilya Sutskever, et al. Language models are few‐shot learners. In NeurIPS, 2020

    work page 2020

  6. [6]

    Campbell, J

    Andrew Campbell, Joe Benton, Valentin De Bortoli, Tom Rainforth, George Deligiannidis, and Arnaud Doucet. A continuous time framework for discrete denoising models, 2022. URL https://arxiv.org/abs/2205.14987

    work page arXiv 2022

  7. [7]

    Trans-dimensional generative modeling via jump diffusion models, 2023

    Andrew Campbell, William Harvey, Christian Weilbach, Valentin De Bortoli, Tom Rainforth, and Arnaud Doucet. Trans-dimensional generative modeling via jump diffusion models, 2023. URL https://arxiv.org/abs/2305.16261

    work page arXiv 2023

  8. [8]

    De Cao and T

    Nicola De Cao and Thomas Kipf. Molgan: An implicit generative model for small molecular graphs, 2022. URL https://arxiv.org/abs/1805.11973

    work page arXiv 2022

  9. [9]

    Itai Gat, Tal Remez, Neta Shaul, Felix Kreuk, Ricky T. Q. Chen, Gabriel Synnaeve, Yossi Adi, and Yaron Lipman. Discrete flow matching, 2024. URL https://arxiv.org/abs/2407.15595

    work page arXiv 2024

  10. [10]

    Bayesian flow networks, 2025

    Alex Graves, Rupesh Kumar Srivastava, Timothy Atkinson, and Faustino Gomez. Bayesian flow networks, 2025. URL https://arxiv.org/abs/2308.07037

    work page arXiv 2025

  11. [11]

    A systematic survey on deep generative models for graph generation, 2022

    Xiaojie Guo and Liang Zhao. A systematic survey on deep generative models for graph generation, 2022. URL https://arxiv.org/abs/2007.06686

    work page arXiv 2022

  12. [12]

    Diffusion models for graphs benefit from discrete state spaces, 2023

    Kilian Konstantin Haefeli, Karolis Martinkus, Nathanaël Perraudin, and Roger Wattenhofer. Diffusion models for graphs benefit from discrete state spaces, 2023. URL https://arxiv.org/abs/2210.01549

    work page arXiv 2023

  13. [13]

    Denoising Diffusion Probabilistic Models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models, 2020. URL https://arxiv.org/abs/2006.11239

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  14. [14]

    Score-based generative modeling of graphs via the system of stochastic differential equations, 2022

    Jaehyeong Jo, Seul Lee, and Sung Ju Hwang. Score-based generative modeling of graphs via the system of stochastic differential equations, 2022. URL https://arxiv.org/abs/2202.02514

    work page arXiv 2022

  15. [15]

    Graph generation with diffusion mixture, 2024

    Jaehyeong Jo, Dongki Kim, and Sung Ju Hwang. Graph generation with diffusion mixture, 2024. URL https://arxiv.org/abs/2302.03596

    work page arXiv 2024

  16. [16]

    Elucidating the Design Space of Diffusion-Based Generative Models

    Tero Karras, Miika Aittala, Timo Aila, and Samuli Laine. Elucidating the design space of diffusion-based generative models, 2022. URL https://arxiv.org/abs/2206.00364

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  17. [17]

    Auto-Encoding Variational Bayes

    Diederik P Kingma and Max Welling. Auto-encoding variational bayes, 2013. URL https://arxiv.org/abs/1312.6114

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  18. [18]

    Variational Graph Auto-Encoders

    Thomas N. Kipf and Max Welling. Variational graph auto-encoders, 2016. URL https://arxiv.org/abs/1611.07308

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  19. [19]

    Hamilton, David Duvenaud, Raquel Urtasun, and Richard S

    Renjie Liao, Yujia Li, Yang Song, Shenlong Wang, Charlie Nash, William L. Hamilton, David Duvenaud, Raquel Urtasun, and Richard S. Zemel. Efficient graph generation with graph recurrent attention networks, 2020. URL https://arxiv.org/abs/1910.00760

    work page arXiv 2020

  20. [20]

    u dke, Jan Hansen-Palmus, and Stephan G \

    Marten Lienen, David L \"u dke, Jan Hansen-Palmus, and Stephan G \"u nnemann. From zero to turbulence: Generative modeling for 3d flow simulation. In ICLR, 2024

    work page 2024

  21. [21]

    Generative modeling with bayesian sample inference, 2025

    Marten Lienen, Marcel Kollovieh, and Stephan Günnemann. Generative modeling with bayesian sample inference, 2025. URL https://arxiv.org/abs/2502.07580

    work page arXiv 2025

  22. [22]

    Graph Normalizing Flows

    Jenny Liu, Aviral Kumar, Jimmy Ba, Jamie Kiros, and Kevin Swersky. Graph normalizing flows, 2019. URL https://arxiv.org/abs/1905.13177

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  23. [23]

    Spectre: Spectral conditioning helps to overcome the expressivity limits of one-shot graph generators, 2022

    Karolis Martinkus, Andreas Loukas, Nathanaël Perraudin, and Roger Wattenhofer. Spectre: Spectral conditioning helps to overcome the expressivity limits of one-shot graph generators, 2022. URL https://arxiv.org/abs/2204.01613

    work page arXiv 2022

  24. [24]

    Permutation invariant graph generation via score-based generative modeling, 2020

    Chenhao Niu, Yang Song, Jiaming Song, Shengjia Zhao, Aditya Grover, and Stefano Ermon. Permutation invariant graph generation via score-based generative modeling, 2020. URL https://arxiv.org/abs/2003.00638

    work page arXiv 2020

  25. [25]

    M olecular S ets ( MOSES ): A B enchmarking P latform for M olecular G eneration M odels

    Daniil Polykovskiy, Alexander Zhebrak, Benjamin Sanchez-Lengeling, Sergey Golovanov, Oktai Tatanov, Stanislav Belyaev, Rauf Kurbanov, Aleksey Artamonov, Vladimir Aladinskiy, Mark Veselov, Artur Kadurin, Simon Johansson, Hongming Chen, Sergey Nikolenko, Alan Aspuru-Guzik, and Alex Zhavoronkov. M olecular S ets ( MOSES ): A B enchmarking P latform for M ole...

    work page 2020

  26. [26]

    Fréchet chemnet distance: A metric for generative models for molecules in drug discovery

    Kristina Preuer, Philipp Renz, Thomas Unterthiner, Sepp Hochreiter, and G \"u nter Klambauer. Fréchet chemnet distance: A metric for generative models for molecules in drug discovery. Journal of Chemical Information and Modeling, 58 0 (9): 0 1736--1741, 2018. doi:10.1021/acs.jcim.8b00234. URL https://doi.org/10.1021/acs.jcim.8b00234. PMID: 30118593

    work page doi:10.1021/acs.jcim.8b00234 2018

  27. [27]

    Defog: Discrete flow matching for graph generation, 2025

    Yiming Qin, Manuel Madeira, Dorina Thanou, and Pascal Frossard. Defog: Discrete flow matching for graph generation, 2025. URL https://arxiv.org/abs/2410.04263

    work page arXiv 2025

  28. [28]

    Malliaros, and Christopher Morris

    Antoine Siraudin, Fragkiskos D. Malliaros, and Christopher Morris. Cometh: A continuous-time discrete-state graph diffusion model, 2024. URL https://arxiv.org/abs/2406.06449

    work page arXiv 2024

  29. [29]

    Deep Unsupervised Learning using Nonequilibrium Thermodynamics

    Jascha Sohl-Dickstein, Eric A. Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynamics, 2015. URL https://arxiv.org/abs/1503.03585

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  30. [30]

    Score-Based Generative Modeling through Stochastic Differential Equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P. Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations, 2021. URL https://arxiv.org/abs/2011.13456

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  31. [31]

    Unified generative modeling of 3d molecules with bayesian flow networks

    Yuxuan Song, Jingjing Gong, Hao Zhou, Mingyue Zheng, Jingjing Liu, and Wei-Ying Ma. Unified generative modeling of 3d molecules with bayesian flow networks. In The Twelfth International Conference on Learning Representations, 2024. URL https://openreview.net/forum?id=NSVtmmzeRB

    work page 2024

  32. [32]

    Smooth interpolation for improved discrete graph generative models

    Yuxuan Song, Juntong Shi, Jingjing Gong, Minkai Xu, Stefano Ermon, Hao Zhou, and Wei-Ying Ma. Smooth interpolation for improved discrete graph generative models. In Forty-second International Conference on Machine Learning, 2025. URL https://openreview.net/forum?id=OYUG5SCg6k

    work page 2025

  33. [33]

    Bayesian flow network framework for chemistry tasks

    Nianze Tao and Minori Abe. Bayesian flow network framework for chemistry tasks. Journal of Chemical Information and Modeling, 65 0 (3): 0 1178--1187, 2025. doi:10.1021/acs.jcim.4c01792

    work page doi:10.1021/acs.jcim.4c01792 2025

  34. [34]

    On the theory of the brownian motion

    George Eugene Uhlenbeck and Leonard Salomon Ornstein. On the theory of the brownian motion. Physical Review, 36 0 (5): 0 823--841, 1930

    work page 1930

  35. [35]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

    work page 2017

  36. [36]

    arXiv preprint arXiv:2209.14734 (2022) 13

    Clement Vignac, Igor Krawczuk, Antoine Siraudin, Bohan Wang, Volkan Cevher, and Pascal Frossard. Digress: Discrete denoising diffusion for graph generation, 2023. URL https://arxiv.org/abs/2209.14734

    work page arXiv 2023

  37. [37]

    Discrete-state continuous-time diffusion for graph generation, 2024

    Zhe Xu, Ruizhong Qiu, Yuzhong Chen, Huiyuan Chen, Xiran Fan, Menghai Pan, Zhichen Zeng, Mahashweta Das, and Hanghang Tong. Discrete-state continuous-time diffusion for graph generation, 2024. URL https://arxiv.org/abs/2405.11416

    work page arXiv 2024

  38. [38]

    Unifying bayesian flow networks and diffusion models through stochastic differential equations

    Kaiwen Xue, Yuhao Zhou, Shen Nie, Xu Min, Xiaolu Zhang, Jun Zhou, and Chongxuan Li. Unifying bayesian flow networks and diffusion models through stochastic differential equations, 2024. URL https://arxiv.org/abs/2404.15766

    work page arXiv 2024

  39. [39]

    GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Models

    Jiaxuan You, Rex Ying, Xiang Ren, William L. Hamilton, and Jure Leskovec. Graphrnn: Generating realistic graphs with deep auto-regressive models, 2018. URL https://arxiv.org/abs/1802.08773

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  40. [40]

    A survey on deep graph generation: Methods and applications, 2022

    Yanqiao Zhu, Yuanqi Du, Yinkai Wang, Yichen Xu, Jieyu Zhang, Qiang Liu, and Shu Wu. A survey on deep graph generation: Methods and applications, 2022. URL https://arxiv.org/abs/2203.06714

    work page arXiv 2022

  41. [41]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should not add it explicitly Type <Return> for now, but then later remove the command n...

    work page

  42. [42]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@first@sw \@firstoftwo \@ifundefined NAT@b*@#2 \@firstoftwo @num @NAT@ctr \@secondoft...

    work page

  43. [43]

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibsetup #1 @NAT@ctr @ @openbib .11em \@plus.33em \@minus.07em 4000 4000 `\.\@m @bibit...

    work page