/L2T_loss

simple demo codes for Learning to Teach with Dynamic Loss Functions

Primary LanguagePython

Learning to Teach with Dynamic Loss Functions

This repo contains the simple demo for the NeurIPS-18 paper: Learning to Teach with Dynamic Loss Functions.

@inproceedings{wu2018learning,
  title={Learning to teach with dynamic loss functions},
  author={Wu, Lijun and Tian, Fei and Xia, Yingce and Fan, Yang and Qin, Tao and Jian-Huang, Lai and Liu, Tie-Yan},
  booktitle={Advances in Neural Information Processing Systems},
  pages={6466--6477},
  year={2018}
}

Description

  • Please note this is only a simple demo for the Mnist experiments based on Lenet.
  • Please note the algorithm in the demo is little different to the paper, but the main spirit is same.
  • The code is based on the Theano framework, which is somehow too old to directly apply this code.

Detailed Critical Codes

Refer to loss_lenet_light_dynamic.py for the detailed demo codes. The general comments are here:

The 'reverse model training' defines the updates of the teacher model, and the last one is the detailed reverse model update chains.