Manqing Ma presented a paper titled "Learning Parameters for Balanced Index Influence Maximization,' co-authored by Gyorgy Korniss and Boleslaw Szymanski at Complex Networks Conference in Madrid, Dec. 3 2020

Manqing Ma presented a paper titles "Learning Parameters for Balanced Index Influence Maximization," co-authored by Gyorgy Korniss and Boleslaw Szymanski published in the Proceedings of the 9th International Conference on Complex Networks and their Applications, December 1-3, 2020, Madrid, Spain (virtual). The paper focuses on influence maximization that is the task of finding the smallest set of nodes whose activation in a social network triggers an activation cascade that reaches the targeted network coverage while using the threshold rules to determine the outcome of influence. This problem is NP-hard and it has generated a significant amount of recent research on finding efficient heuristics. The authors focus on a Balance Index algorithm that relies on three parameters to tune its performance to the given network structure. Using a supervised machine-learning approach, he authors select the most influential graph features for the parameter tuning. Then, using random-walk-based graph-sampling, small snap-shots are created from the given synthetic and large-scale real-world networks. Using exhaustive search on these snapshots, the high accuracy values of BI parameters are found are used as a ground truth. Then, a machine-learning model is trained on the snapshots and applied to the real-word network to find the best BI parameters. The found parameters are applied to the sampled real-world network to measure the quality of the sets of initiators found this way. The authors use various real-world networks to validate this approach against other heuristic.