Search Strategies for Topological Network Optimization

Authors: Michael D. Moffitt10299-10308

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we evaluate our approach on a new parameterized benchmark suite, demonstrating over an order of magnitude improvement in performance as compared to a baseline implementation.
Researcher Affiliation Industry Michael D. Moffitt Google moffitt@google.com
Pseudocode Yes Algorithm 1: NETWORK-OPT
Open Source Code Yes Source code (including visualization utilities) is available at https://github.com/google/network-opt
Open Datasets No The paper mentions a parameterized benchmark suite based on the E12 series of preferred values, which is a standard, but it does not provide a direct link, DOI, or specific repository for a dataset file. It describes how the values for S and fT are assigned based on this standard.
Dataset Splits No The paper does not explicitly mention specific train/validation/test splits, sample counts for splits, or cross-validation setup. It evaluates performance on parameterized problems of increasing 'n'.
Hardware Specification Yes All experiments were conducted on a Debian Linux workstation powered by a 2.20GHz Intel Xeon CPU and 32gb of RAM.
Software Dependencies No The paper mentions the operating system "Debian Linux" but does not specify versions for any other software dependencies such as programming languages, libraries, or frameworks.
Experiment Setup Yes A timeout of one week was imposed on all solvers. For any experiment using tabulation, the hyperparameter M was set to n/2 .