[deep learning] multi task learning of multiple datasets data sets missing labels

Data sets 1 Is the text emotion classification task data set , Data sets 2 Is a text subject classification dataset , The model is the same backbone, Can multiple datasets be applied to such a model of multiple tasks , If there is a good theoretical support , There will be 2 Advantages ：
（1） A lot of multi task learning is about the same data set applied to different tasks , There is... In the middle of the task “ The lines ” The relationship between , It is conducive to the training of the model . Use the same... In this problem backbone, Will still get “ The lines ” The relationship between .
（2） Different datasets need not be labeled separately . Data missing mark problem .

You can see from these directions ：
（1）Multi-Task Deep Neural Networks for Natural Language Understanding
（2）https://github.com/xingyizhou/UniDet

Reference resources ：
Introduction to multi task learning in deep learning
Multi-Task Deep Neural Networks for Natural Language Understanding
Can I input multiple data sets and train multiple tasks at the same time
https://github.com/xingyizhou/UniDet