The modules in this directory implement:
- DEC (Unsupervised Deep Embedding for Clustering Analysis) and
- IDEC (Xifeng Guo, Long Gao, Xinwang Liu, Jianping Yin. Improved Deep Embedded Clustering with Local Structure Preservation. IJCAI 2017)
Everything is written in pure
python on only the MNIST dataset. Special thanks to the authors for providing a Keras implementation here, which we base a lot of this on. Testing error in our implementation converges between 87-88%.
The key differences between our implementation and the paper are as follows:
- We train our autoencoder end-to-end; that is, we do not do greedy, layer-wise training.
- We used Adam as the optimizer for both pretraining and training.