$\texttt{LeadCache}$: Regret-Optimal Caching in Networks
Authors: Debjit Paria, Abhishek Sinha
NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 7 Experiments In this section, we compare the performance of the Lead Cache policy with standard caching policies. Figure 3 shows the performance of different caching policies in terms of the average cache hit rate per file and the average number of file fetches per cache. |
| Researcher Affiliation | Academia | Debjit Paria Department of Computer Science Chennai Mathematical Institute Chennai 603103, India Abhishek Sinha Department of Electrical Engineering Indian Institute of Technology Madras Chennai 600036, India |
| Pseudocode | Yes | Algorithm 1 The Lead Cache Policy Algorithm 2 Cache-wise Deterministic Pipage rounding Algorithm 3 Randomized Rounding with Madow s Sampling Scheme |
| Open Source Code | Yes | The code for the experiments is available online [Paria and Sinha, 2021]. Source code for Lead Cache: Regret-Optimal Caching in Networks. https://github.com/AbhishekMITIITM/LeadCache-NeurIPS21, 2021. |
| Open Datasets | Yes | In our experiments, we use a publicly available anonymized production trace from a large CDN provider available under a BSD 2-Clause License [Berger et al., 2018, Berger, 2018]. |
| Dataset Splits | No | The paper states: 'We divide the trace consisting of the first 375K requests into 20 consecutive sub-intervals.' This describes a division for processing, but does not specify a training/validation/test split for machine learning model development in terms of percentages or sample counts. |
| Hardware Specification | No | The computational work reported on this paper was performed on the AQUA Cluster at the High Performance Computing Environment of IIT Madras. No specific hardware details such as CPU or GPU models, or memory sizes, are provided. |
| Software Dependencies | No | The paper does not explicitly list software dependencies with specific version numbers (e.g., Python, specific libraries, or frameworks with versions). |
| Experiment Setup | Yes | In our experiments, we use a publicly available anonymized production trace from a large CDN provider... We construct a random Bipartite caching network with n = 30 users and m = 10 caches. Each cache is connected to d = 8 randomly chosen users. ... The storage capacity C of each cache is taken to be 10% of the catalog size. We divide the trace consisting of the first 375K requests into 20 consecutive sub-intervals. |