Cost-Sensitive Multi-Label Classification

Python implementation of our paper Cost-Sensitive Label Embedding for Multi-Label Classification and related algorithms, including:

• Cost-Sensitive Label Embedding with Multidimensional Scaling (CLEMS)
• Condensed Filter Tree (CFT)
• Probabilistic Classifier Chains (PCC)
• Classifier Chains (CC)
• Binary Relevance (BR)

If you find our paper or implementation is useful in your research, please consider citing our paper for CLEMS and the references below for other algorithms.

@article{Huang2017clems,
    author    = {Kuan-Hao Huang and
                 Hsuan-Tien Lin},
    title     = {Cost-sensitive label embedding for multi-label classification},
    journal   = {Machine Learning},
    volume    = {106},
    number    = {9-10},
    pages     = {1725--1746},
    year      = {2017},
}

Prerequisites

• Python 2.7.12
• NumPy 1.13.3
• scikit-learn 0.17

Usage

$ python demo.py

Dataset

• scene (downloaded from Mulan)

Evaluation Criteria

• Hamming loss
• Rank loss
• F1 score
• Accuracy score

Result

============================================================
algorithm  hamming_loss  rank_loss  f1_score  accuracy_score
============================================================
       BR        0.0907     1.1844    0.5742          0.5627
       CC        0.0880     1.1424    0.5947          0.5851
      PCC        0.0900     0.6898    0.7460          0.6909
      CFT        0.0867     0.9460    0.6478          0.6267
    CLEMS        0.0825     0.6553    0.7690          0.7600
============================================================

Reference

• Grigorios Tsoumakas and Ioannis Katakis. Multi-Label Classification: An Overview. International Journal of Data Warehousing and Mining, 2007.

• Jesse Read, Bernhard, Pfahringer, Geoff Holmes, and Eibe Frank. Classifier chains for multi-label classification. Machine Learning, 2011

• Krzysztof Dembczynski, Weiwei Cheng, and Eyke Hullermeier. Bayes Optimal Multilabel Classification via Probabilistic Classifier Chains. ICML, 2012.

• Chun-Liang Li and Hsuan-Tien Lin. Condensed Filter Tree for Cost-Sensitive Multi-Label Classification. ICML, 2014.

• Kuan-Hao Huang and Hsuan-Tien Lin. Cost-Sensitive Label Embedding for Multi-Label Classification. Machine Learning, 2017

Author

Kuan-Hao Huang / @ej0cl6