morning-dews/PGMA_tensorflow

Code for Overcoming Catastrophic Forgetting for Continual Learning via Model Adaptation

Parameter Generation and Model Adaptation(PGMA)

This is an implementation of Overcoming Catastrophic Forgetting for Continual Learning via Model Adaptation(Hu et al.,2018) in disjoint MNIST setting.

Requirements

• Python 3.5
• tensorflow 1.4

Explanations

wae.py is based on the implementation of WAE. It defines the network structure and training procedure of our model.

configs.py defines the configurations in our model.

setdata.py loads MNIST data.

run.py is the main script to train and test our model.

ops.py and models.py define operations and functions utilized in wae.py.