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..
Controlling The Spread of Epidemics on Networks with Differential Privacy
Authors: Dũng Nguyen, Aravind Srinivasan, Renata Valieva, Anil Vullikanti, Jiayi Wu
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate the performance of our algorithms on different realistic and random networks in terms of the following questions Effects of privacy budgets on the utility of our algorithm (both in terms of vaccination budget and epidemic metrics (G) and ρ(G)). Tradeoff between vaccination cost, different epidemic metrics, and privacy parameters. Comparison between the implicit and explicit solutions. ... Figure 1: Effect of Privacy on Budget Requirements on Montgomery County Subnets ... Table 1: Network datasets used in evaluation |
| Researcher Affiliation | Academia | Dung Nguyen Department of Computer Science Haverford College Haverford, PA 19041 EMAIL Aravind Srinivasan Department of Computer Science University of Maryland College Park, MD 20742 EMAIL Renata Valieva Department of Mathematics University of Maryland College Park, MD 20742 EMAIL Anil Vullikanti Department of Computer Science University of Virginia Charlottesville, VA 22904 EMAIL Jiayi Wu Department of Computer Science University of Maryland College Park, MD 20742 EMAIL |
| Pseudocode | Yes | Algorithm 1 Private algorithm for PRIVATEMAXDEG ... Algorithm 2 Explicit solution algorithm for PRIVATEMAXDEG ... Algorithm 3 Private Hitting Walks Algorithm for PRIVMINSR ... Algorithm 4 Private algorithm for PRIVATEMAXDEG ... Algorithm 5 Private algorithm for PRIVATEMULSET ... Algorithm 6 Private algorithm for WEIGHTEDPRIVATEMULSET |
| Open Source Code | No | 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 instructions to reproduce the experiments are detailed in the algorithms. All datasets are publicly available with their sources cited. |
| Open Datasets | Yes | We consider two classes of networks, as summarized in Table 1. The digital twin of a contact network [2, 16] is a model of real world activity based contact networks; we consider three subgraphs with 10,000 nodes of the network for Montgomery county VA. The BTER model [28] is a random graph model, which preserves both degree sequence and clustering; we consider three randomly generated networks. Both classes of networks have been used in a number of epidemiological analyses, e.g., [32, 8, 1]. ... Finally, we computed the 300 explicit solutions using various privacy budgets ϵ (and δ = 0.01) and target max degree 10 for 3 social circles in the SNAP Facebook datasets [29] |
| Dataset Splits | No | The paper lists several datasets (Montgomery County subgraphs, BTER networks, SNAP Facebook datasets) and describes their general use in experiments and simulations, but it does not specify any particular training, testing, or validation splits or percentages for these datasets. |
| Hardware Specification | No | Question: For each experiment, does the paper provide sufficient information on the computer resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [No] Justification: We presented the computational complexity of our algorithms, which was quite reasonable for the networks we considered in our experiments. Since the experiments can be performed by typical workstations without specialized hardware and in relatively short amount of time, we have decided not to include such information. |
| Software Dependencies | No | The paper does not explicitly mention any specific software components (libraries, frameworks, or programming languages) along with their version numbers that were used in the implementation or experimentation. |
| Experiment Setup | Yes | We use a privacy budget of δ = 10^-6 and ϵ {0.25, 0.5, 1, 2, 4}, and set a target degree of D = 45. ... We pick δ = 1/n = 10^-3 here, and have relaxed the privacy to the multi-set multi-cover definition rather than the edge private definition of neighboring datasets. ... Finally, we computed the 300 explicit solutions using various privacy budgets ϵ (and δ = 0.01) and target max degree 10 for 3 social circles in the SNAP Facebook datasets [29], we then performed 200 simulations of SIR with transmission probability 0.2 and 20 initial infections |