t-sakai-kure / pywsl Goto Github PK

View Code? Open in Web Editor NEW

122.0 7.0 29.0 13.09 MB

Python codes for weakly-supervised learning

License: MIT License

Python 100.00%

machine-learning python semi-supervised-learning positive-unlabeled-learning classification unlabeled-data

pywsl's Introduction

pywsl: python codes for weakly-supervised learning

This package contains the following implementation:

Unbiased PU learning
in "Convex formulation for learning from positive and unlabeled data", ICML, 2015 [uPU]
Non-negative PU Learning
in "Positive-unlabeled learning with non-negative risk estimator", NIPS, 2017 [nnPU]
PU Set Kernel Classifier
in "Convex formulation of multiple instance learning from positive and unlabeled bags", Neural Networks, 2018 [PU-SKC]
Class-prior estimation based on energy distance
in "Computationally efficient class-prior estimation under class balance change using energy distance", IEICE-ED, 2016 [CPE-ENE].
PNU classification
in "Semi-supervised classification based on classification from positive and unlabeled data", ICML 2017 [PNU].
PNU-AUC optimization
in "Semi-supervised AUC optimization based on positive-unlabeled learning", MLJ 2018 [PNU-AUC].

Installation

$ pip install pywsl

Main contributors

References

du Plessis, M. C., Niu, G., and Sugiyama, M. Convex formulation for learning from positive and unlabeled data.
In Bach, F. and Blei, D. (Eds.), Proceedings of 32nd International Conference on Machine Learning, JMLR Workshop and Conference Proceedings, vol.37, pp.1386-1394, Lille, France, Jul. 6-11, 2015.
Kiryo, R., Niu, G., du Plessis, M. C., and Sugiyama, M.
Positive-unlabeled learning with non-negative risk estimator.
In Guyon, I., Luxburg, U. V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., and Garnett, R. (Eds.), Advances in Neural Information Processing Systems 30, pp.1674-1684, 2017.
Bao, H., Sakai, T., Sato, I., and Sugiyama, M.
Convex formulation of multiple instance learning from positive and unlabeled bags.
Neural Networks, vol.105, pp.132-141, 2018.
Kawakubo, H., du Plessis, M. C., and Sugiyama, M.
Computationally efficient class-prior estimation under class balance change using energy distance.
IEICE Transactions on Information and Systems, vol.E99-D, no.1, pp.176-186, 2016.
Sakai, T., du Plessis, M. C., Niu, G., and Sugiyama, M.
Semi-supervised classification based on classification from positive and unlabeled data.
In Precup, D. and Teh, Y. W. (Eds.), Proceedings of 34th International Conference on Machine Learning, Proceedings of Machine Learning Research, vol.70, pp.2998-3006, Sydney, Australia, Aug. 6-12, 2017.
Sakai, T., Niu, G., and Sugiyama, M.
Semi-supervised AUC optimization based on positive-unlabeled learning.
Machine Learning, vol.107, no.4, pp.767-794, 2018.

pywsl's People

Contributors

Stargazers

Watchers

pywsl's Issues

New feature: PU-AUC optimization

fetch_mldata Deprecated and Down Routinely

I was looking at your nnPU chainer implementation. If I am not mistaken, it seems to download data from mldata.org. This sklearn issue report discusses the instability of mldata.org and how it often goes down for extended periods of time. For example, mldata.org is down at the moment.

When you run the fetch_mldata command, you get the warning in sklearn:

DeprecationWarning: Function fetch_mldata is deprecated; fetch_mldata was deprecated in version 0.20 and will be removed in version 0.22

What are your thoughts on moving away from fetch_mldata to an alternative? Different alternatives include:

Hosting the file on Github (probably in a different repository)
Moving to another source for the data, e.g., torchvision

I would be happy to work with you on this if you are interested.

Feature Request: Use PNU classification with a scikit-learn cross validator

Hello.

Currently, PNU_SL function internally computes the cross validation score. I'd like to use PNU classification with a scikit-learn cross validator. Is it possible to implement PNU_SL class conforming to scikit-learn?

Thank you.

Issues running the demo file

PNU/python/cpe_ene.py:49: RuntimeWarning: invalid value encountered in sqrt
  return np.sqrt(dist)
PNU/python/pnu_sl.py:152: RuntimeWarning: invalid value encountered in greater_equal
  f_p = np.mean(g_n >= 0) if n_n != 0 else 0
PNU/python/pnu_sl.py:161: RuntimeWarning: invalid value encountered in less_equal
  f_nu = np.mean(g_u <= 0) if n_u != 0 else 0
Traceback (most recent call last):
  File "./demo.py", line 64, in <module>
    model='lm', nargout=3)
  File "PNU/python/pnu_sl.py", line 92,in PNU_SL
    sigma, lam = sigma_list[best_sigma_index], lambda_list[best_lambda_index]
UnboundLocalError: local variable 'best_sigma_index' referenced before assignment

Hello, I've been trying to run the demo file, and encountered an error presented above.
any way to fix this?
Thanks in advance.