Dynamic selection of a video content adaptation strategy from a Pareto front

Abstract

Genetic Algorithms may be used together with Pareto Optimality in the process of selection of a suitable video content adaptation strategy, the former to return best or fittest solutions that have evolved over many generations and the latter to evaluate and rank each generation's solutions against a set of objectives without the need to assign weights to each one. The outcome of this is a Pareto front of optimal strategies, all of which would satisfy the objectives. The distribution of optimal strategies on a Pareto front, however, suggests that there may be a ‘best-fit’ optimal strategy. This article refines the process of selection of an optimal strategy by taking into account this distribution alongside user preferences, video content characteristics and usage history. In order to make the refined process dynamic, it pursues its implementation using Self-Organising Neural Networks.