RA-MMOE

1.运用场景

    CTR模型/多任务学习。

2.创新点

    propose a novel multi-task learning approach,multi-gate Mixture-of-Experts,which explicitly learns to model task relationships from data.
    adapt the Mixture-of-Experts structure to multi-task learning by sharing the expert submodels across all task,while also having a gating network trained to optimize each task.

3.算法原理

3.1 网络框架

3.2 MMOE

    MMOE论文

4.算法理解

    多任务学习，通用网络。