MENDR

Measurement Error in Network Diffusion & Reconstruction

MENDR is a benchmarking data and results registry, along with tooling for accurate (de)serialization and validation of each challenge problem.

Installing

mendr is currently un-published. Reference installations can be achieved for development purposes with

pip install git+https://github.com/usnistgov/mendr.git

Dataset Overview

The MENDR dataset contains several thousand challenge datasets, where the goal is to recover a graph structure from a set of random-walk visitations. If a node is considered "active" when a random walk has visited it, we can create a binary node activation vector from the set of visited nodes. Then, running many random walks on a graph will produce a binary node "activation" matrix.

Every graph+activation pair has a unique ID in MENDR. The id will start with its code:

BL : block network

TR : Tree network

SC : Scale-free (Barabasi-Albert)

This is followed by a code containing the number of nodes and the seed that generated the random sample, e.g. BL-N030S01.

Other parameters are sampled randomly for each combination of the above, as detailed in the folowing table:

parameters	values
random graph kind	Tree, Block, BA$(m\in{1,2}$)
network $n$-nodes	10,30,100,300
random walks	1 sample $m\sim\text{NegBinomial}(2,\tfrac{1}{n})+10$
random walk jumps	1 sample $j\sim\text{Geometric}(\tfrac{1}{n})+5$
random walk root	1 sample $n_0 \sim \text{Multinomial}(\textbf{n},1)$
random seed	1, 2, ... , 30

: Experiment Settings (MENDR Dataset) {#tbl-mendr}