This is a wrapper script for the BPMLL algorithm [1, 2]. It runs this model on 4 benchmark multilabel datasets and outputs evaluation metrics.

Requirements

Matlab
Neural Network Toolkit for Matlab. (to check if you have it, type 'help newff' and see if you get a function description)

How to run

Download this repo and navigate to the directory that contains this readme file.
```
  git clone https://github.com/abhishek-kumar/BPMLL
  cd BPMLL
```
copy the 'data.zip' file from https://www.dropbox.com/s/g812jg9c6kj081y/data.zip to the current directory
extract 'data.zip' here. After extracting, the 'data' directory should now contain four subdirectories, 'scene', 'yeast', 'emotions' and 'enron'.
Start Matlab, navigate to the current directory (BPMLL), and type 'run'

Note

After extracting 'data.zip', you should have the following directory structure within the folder 'data':

scene/scene-train.csv
scene/scene-test.csv
yeast/yeast-train.csv
yeast/yeast-test.csv
emotions/emotions-train.csv
emotions/emotions-test.csv
enron/enron-train.csv
enron/enron-test.csv

Dataset format. Each dataset is separated into two CSV files - a training set and a test set. If the training set contains 4 examples with 3 features and 2 labels, the train.csv file should have 4 rows and 3+2=5 columns. Labels must be in {0,1} and feature values must be in [-1,1]. E.g.
```
 0.5677,0.7400,0.0765,0,1
 0.8364,-0.7465,-0.9874,1,1
 0.7465,0.0000,0.9048,1,0
 0.8475,-0.8475,0.3658,0,0
```

References

[1] M.-L. Zhang and Z.-H. Zhou. Multilabel neural networks with applications to functional genomics and text categorization. IEEE Transactions on Knowledge and Data Engineering, 2006, 18(10): 1338-1351.

[2] The BPMLL source code within the directory 'BPMLL-src' is taken from http://lamda.nju.edu.cn/Default.aspx?Page=code_BPMLL

abhishek-kumar/BPMLL

Requirements

How to run

Note

References