Soft parameter sharing is the regularization between parameter distance to make the parameters similar. However, in soft parameter sharing each task has its own parameters.

University of Michigan - Ann Arbor

1) Hard parameter sharing
2) Soft parameter sharing


Two types of Multi task learning methods


Soft Parameter Sharing


Hard parameter sharing is a multi-task learning (MTL) approach for neural networks in which the hidden layers are shared across all tasks, while each task keeps its own task-specific output layers. Because the shared layers must fit patterns common to every task rather than the idiosyncrasies of any single task, this approach reduces the risk of overfitting compared to training a separate network per task.

Learn Before

Related