The performance evaluation of Gossip Learning protocols when we have dynamic setups
|Author||Mina AGHAEI DINANI|
|Director of thesis|
|Co-director of thesis|
|Summary of thesis||
In our approaches, nodes improve their personalized model instance by sharing it with neighbors, and by weighting neighbors’ contributions according to an estimate of their marginal utility. Then we apply our GL algorithms on different datasets to evaluate their performance.
|Administrative delay for the defence|