A clustering strategy with deep neural networks. This blog article provides a generic overview.

This repository provides the PyTorch implementation of the transfer learning schemes (L2C) and two deep clustering criteria:

• KLD-based Contrastive Loss (KCL)
• Constrained Clustering Likelihood (CCL)

pip install -r requirements.txt

# A quick trial:
python demo.py  # Dataset:MNIST, Network:LeNet, Loss:CCL
python demo.py --loss KCL

# Lookup available options:
python demo.py -h

# For more examples:
./scripts/exp_supervised_CCL_vs_KCL.sh

# Learn the Similarity Prediction Network (SPN) with Omniglot_background and then transfer to the 20 alphabets in Omniglot_evaluation.
# Default loss is CCL with an unknown number of clusters (Set a large cluster number, i.e., k=100)
# It takes about half an hour to finish.
python demo_omniglot_transfer.py

# An example of using KCL and set k=gt_#cluster
python demo_omniglot_transfer.py --loss KCL --num_cluster -1

# Lookup available options:
python demo_omniglot_transfer.py -h

# Other examples:
./scripts/exp_unsupervised_transfer_Omniglot.sh

• The clustering results are highly dependent on the performance of the Similarity Prediction Network (SPN). For making a fair comparison, the SPN must be kept the same. Our script trains an SPN with random initialization and random data sampling. Once the SPN model is trained, the script will reuse the saved SPN and avoid training a new one.
• The table below presents the clustering performance with the reference SPN [download]. Put the model file into /outputs folder and run demo_omniglot_transfer.py directly to generate the "CCL(k=100)" column.
• The performance metric is clustering accuracy (for details, please see L2C paper). Each value in the table is the average of 3 clustering runs. This repository reuses most of the utilities in PyTorch and is different from the Lua-based implementation used in the reference papers. The result (the row with "--Average--") shows the same trend as the papers, but the absolute values have a mild difference. The CCL results here are better than the report.

@article{Hsu18_L2C,
	title =     {Learning to cluster in order to transfer across domains and tasks},
	author =    {Yen-Chang Hsu and Zhaoyang Lv and Zsolt Kira},
	booktitle = {International Conference on Learning Representations (ICLR)},
	year =      {2018},
	url =       {https://openreview.net/forum?id=ByRWCqvT-}
}

@article{Hsu18_CCL,
	title =	    {A probabilistic constrained clustering for transfer learning and image category discovery},
	author =    {Yen-Chang Hsu, Zhaoyang Lv, Joel Schlosser, Phillip Odom, Zsolt Kira},
	booktitle = {CVPR Deep-Vision workshop},
	year =      {2018},
	url =       {https://arxiv.org/abs/1806.11078}
}

@article{Hsu16_KCL,
	title =	    {Neural network-based clustering using pairwise constraints},
	author =    {Yen-Chang Hsu and Zsolt Kira},
	booktitle = {ICLR workshop},
	year =      {2016},
	url =       {https://arxiv.org/abs/1511.06321}
}

@article{Hsu18_InsSeg,
	title =     {Learning to Cluster for Proposal-Free Instance Segmentation},
	author =    {Yen-Chang Hsu, Zheng Xu, Zsolt Kira, Jiawei Huang},
	booktitle = {accepted to the International Joint Conference on Neural Networks (IJCNN)},
	year =      {2018},
	url =       {https://arxiv.org/abs/1803.06459}
}

This work was supported by the National Science Foundation and National Robotics Initiative (grant # IIS-1426998).