This is DeepCluster for chainer. But this implementation is not fixed.
The paper of this wonderful model is below.
Deep Clustering for Unsupervised Learning of Visual Features
- Evaluator
- Feature that select your data
- Model implementation
- hyper-parameter tuning
- Prepare the Demo
$ python train.py [dataset_name] [--output_dir 'result'] [--batchsize 256] [--epoch 300] [--gpu 0] [--fully_output_size 100] [--pca_dim 128] [--verbose] [--gpu]
for example:
$ python train.py mnist --output_dir='result/mnist/' --epoch 50
support dataset is bellow:
- mnist
- fashion-mnist
- CIFAR 10
- CIFAR 100