Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Diffusion Network Inference for Cross-layer Cascades
Authors: Siyu Huang, Yubai Yuan, Abdul Basit Adeel
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 4 Numerical experiments on synthetic cascading data We investigate the performance of the proposed method in recovering the diffusion networks based on synthetic cascading data. 5 Real cascading data analysis In this section, we study the cascading patterns of research topics in sociology by discovering the diffusion networks among US universities. |
| Researcher Affiliation | Academia | Siyu Huang Department of Statistics The Pennsylvania State University EMAIL Yubai Yuan : Department of Statistics The Pennsylvania State University EMAIL Abdul Basit Adeel Department of Sociology and Criminology The Pennsylvania State University EMAIL |
| Pseudocode | Yes | Algorithm 1 First-order projected EM algorithm Require: initialization Ωp0q tΘp0q, Ψp0q, πp0qu, observed network A, low-rank penalty µ, learning rate λ, and stopping criterion ϵ. while QpΩps 1q | Ωpsqq µ}Ψps 1q} QpΩpsq | Ωpsqq µ}Ψpsq} ą ϵ do. E-step: update ˆπc i P p Zi | tc; Ωpsqq via its posterior distribution based on Ωpsq for i 1, , N, c 1, , C. M-step: decompose QpΩ| Ωpsqq Q1pΘ | Ωpsqq Q2pΨ | Ωpsqq Q3pπ | Ωpsqq M.1: Update Θ via Q1pΘ | Ωpsqq: Θps 1q Ð maxtΘpsq λ BQ1 Θ ˇˇˇ Θ Θpsq d A, 0u M.2: Update Ψ via Q2pΨ | Ωpsqq: M.2.1: Ψ Ð Ψ λ BQ2 Ψ ˇˇˇ Ψ Ψpsq M.2.2: perform SVD decomposition on Ψ UΛV J M.2.3: Ψps 1q Ð maxt UdiagpΛ λµq V , 0u M.3: Update π via Q3pπ | Ωpsqq: řC c 1 ˆπc i C , i 1, , N end while |
| Open Source Code | Yes | 5. Open access to data and code Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: The paper provides open access to the data and code with sufficient instructions to faithfully reproduce the main experimental results. 13. New assets Question: Are new assets introduced in the paper well documented and is the documentation provided alongside the assets? Answer: [Yes] Justification: New code accompanying this paper is well documented and submitted as supplementary material. |
| Open Datasets | Yes | 5. Open access to data and code Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: The paper provides open access to the data and code with sufficient instructions to faithfully reproduce the main experimental results. |
| Dataset Splits | No | Likelihood-based parameter tuning The low-rank penalty µ in the above Algorithm 1 can be selected in a data-adapted manner. Specifically, we can first randomly separate the total cascade samples into a training subset Ctrain and a validation subset Cval, and estimate model parameters ˆΩµ on Ctrain given a specific µ. |
| Hardware Specification | Yes | Numerical experiments in this paper were carried out in Google Colab on a single Nvidia A100 GPU with 80GB of memory available. |
| Software Dependencies | Yes | In addition, to evaluate the benchmark algorithm Con NIe, we used SNOPT (version 7.7) as the underlying constrained optimization solver [15, 16]. |
| Experiment Setup | Yes | In the following numerical experiments, we fix the size of diffusion networks at N 200. ... Given Θ and Ψ, we generate C 2, 000 independent cascade samples based on double mixture model with Exp transmission model and observation window length being T 10. |