Home

Awesome

<h1 align="center"> MESA: Meta-sampler for imbalanced learning </h1> <p align="center"> <!-- <img src="https://img.shields.io/badge/ZhiningLiu1998-MESA-orange"> <img src="https://img.shields.io/github/stars/ZhiningLiu1998/mesa"> <img src="https://img.shields.io/github/forks/ZhiningLiu1998/mesa"> <img src="https://img.shields.io/github/issues/ZhiningLiu1998/mesa"> <img src="https://img.shields.io/github/license/ZhiningLiu1998/mesa"> --> <a href="https://github.com/ZhiningLiu1998/mesa"> <img src="https://img.shields.io/badge/ZhiningLiu1998-MESA-orange"> </a> <a href="https://github.com/ZhiningLiu1998/mesa/stargazers"> <img src="https://img.shields.io/github/stars/ZhiningLiu1998/mesa"> </a> <a href="https://github.com/ZhiningLiu1998/mesa/network/members"> <img src="https://img.shields.io/github/forks/ZhiningLiu1998/mesa"> </a> <a href="https://github.com/ZhiningLiu1998/mesa/issues"> <img src="https://img.shields.io/github/issues/ZhiningLiu1998/mesa"> </a> <a href="https://github.com/ZhiningLiu1998/mesa/graphs/traffic"> <img src="https://visitor-badge.glitch.me/badge?page_id=ZhiningLiu1998.mesa"> </a> <!-- ALL-CONTRIBUTORS-BADGE:START - Do not remove or modify this section --> <a href="https://github.com/ZhiningLiu1998/mesa#contributors-"><img src="https://img.shields.io/badge/all_contributors-1-orange.svg"></a> <!-- ALL-CONTRIBUTORS-BADGE:END --> <a href="https://github.com/ZhiningLiu1998/mesa/blob/master/LICENSE"> <img src="https://img.shields.io/github/license/ZhiningLiu1998/mesa"> </a> </p> <h3 align="center"> MESA: Boost Ensemble Imbalanced Learning with MEta-SAmpler (NeurIPS 2020) </h3> <h3 align="center"> Links: <a href="https://papers.nips.cc/paper/2020/file/a64bd53139f71961c5c31a9af03d775e-Paper.pdf">Paper</a> | <a href="https://arxiv.org/pdf/2010.08830.pdf">PDF with Appendix</a> | <a href="https://studio.slideslive.com/web_recorder/share/20201020T134559Z__NeurIPS_posters__17343__mesa-effective-ensemble-imbal?s=d3745afc-cfcf-4d60-9f34-63d3d811b55f">Video</a> | <a href="https://arxiv.org/abs/2010.08830">arXiv</a> | <a href="https://zhuanlan.zhihu.com/p/268539195">Zhihu/知乎</a> </h3>

MESA is a meta-learning-based ensemble learning framework for solving class-imbalanced learning problems. It is a task-agnostic general-purpose solution that is able to boost most of the existing machine learning models' performance on imbalanced data.

<!-- > **NOTE:** The paper will be available through [arXiv](https://arxiv.org/) in a few days. We will provide a link to the .pdf file ASAP. -->

Cite Us

If you find this repository helpful in your work or research, we would greatly appreciate citations to the following paper:

@inproceedings{liu2020mesa,
    title={MESA: Boost Ensemble Imbalanced Learning with MEta-SAmpler},
    author={Liu, Zhining and Wei, Pengfei and Jiang, Jing and Cao, Wei and Bian, Jiang and Chang, Yi},
    booktitle={Conference on Neural Information Processing Systems},
    year={2020},
}

Table of Contents

Background

About MESA

We introduce a novel ensemble imbalanced learning (EIL) framework named MESA. It adaptively resamples the training set in iterations to get multiple classifiers and forms a cascade ensemble model. MESA directly learns a parameterized sampling strategy (i.e., meta-sampler) from data to optimize the final metric beyond following random heuristics. It consists of three parts: meta sampling as well as ensemble training to build ensemble classifiers, and meta-training to optimize the meta-sampler.

The figure below gives an overview of the MESA framework.

image

Pros and Cons of MESA

Here are some personal thoughts on the advantages and disadvantages of MESA. More discussions are welcome!

Pros:

Cons:

Requirements

Main dependencies:

To install requirements, run:

pip install -r requirements.txt

NOTE: this implementation requires an old version of PyTorch (v1.0.0). You may want to start a new conda environment to run our code. The step-by-step guide is as follows (using torch-cpu for an example):

These commands should help you to get ready for running mesa. If you have any further questions, please feel free to open an issue or drop me an email.

Usage

A typical usage example:

# load dataset & prepare environment
args = parser.parse_args()
rater = Rater(args.metric)
X_train, y_train, X_valid, y_valid, X_test, y_test = load_dataset(args.dataset)
base_estimator = DecisionTreeClassifier()

# meta-training
mesa = Mesa(
    args=args, 
    base_estimator=base_estimator, 
    n_estimators=10)
mesa.meta_fit(X_train, y_train, X_valid, y_valid, X_test, y_test)

# ensemble training
mesa.fit(X_train, y_train, X_valid, y_valid)

# evaluate
y_pred_test = mesa.predict_proba(X_test)[:, 1]
score = rater.score(y_test, y_pred_test)

Running main.py

Here is an example:

python main.py --dataset Mammo --meta_verbose 10 --update_steps 1000

You can get help with arguments by running:

python main.py --help
optional arguments:
  # Soft Actor-critic Arguments
  -h, --help            show this help message and exit
  --env-name ENV_NAME
  --policy POLICY       Policy Type: Gaussian | Deterministic (default:
                        Gaussian)
  --eval EVAL           Evaluates a policy every 10 episode (default:
                        True)
  --gamma G             discount factor for reward (default: 0.99)
  --tau G               target smoothing coefficient(τ) (default: 0.01)
  --lr G                learning rate (default: 0.001)
  --lr_decay_steps N    step_size of StepLR learning rate decay scheduler
                        (default: 10)
  --lr_decay_gamma N    gamma of StepLR learning rate decay scheduler
                        (default: 0.99)
  --alpha G             Temperature parameter α determines the relative
                        importance of the entropy term against the reward
                        (default: 0.1)
  --automatic_entropy_tuning G
                        Automaically adjust α (default: False)
  --seed N              random seed (default: None)
  --batch_size N        batch size (default: 64)
  --hidden_size N       hidden size (default: 50)
  --updates_per_step N  model updates per simulator step (default: 1)
  --update_steps N      maximum number of steps (default: 1000)
  --start_steps N       Steps sampling random actions (default: 500)
  --target_update_interval N
                        Value target update per no. of updates per step
                        (default: 1)
  --replay_size N       size of replay buffer (default: 1000)

  # Mesa Arguments
  --cuda                run on CUDA (default: False)
  --dataset N           the dataset used for meta-training (default: Mammo)
  --metric N            the metric used for evaluate (default: aucprc)
  --reward_coefficient N
  --num_bins N          number of bins (default: 5). state-size = 2 *
                        num_bins.
  --sigma N             sigma of the Gaussian function used in meta-sampling
                        (default: 0.2)
  --max_estimators N    maximum number of base estimators in each meta-
                        training episode (default: 10)
  --meta_verbose N      number of episodes between verbose outputs. If 'full'
                        print log for each base estimator (default: 10)
  --meta_verbose_mean_episodes N
                        number of episodes used for compute latest mean score
                        in verbose outputs.
  --verbose N           enable verbose when ensemble fit (default: False)
  --random_state N      random_state (default: None)
  --train_ir N          imbalance ratio of the training set after meta-
                        sampling (default: 1)
  --train_ratio N       the ratio of the data used in meta-training. set
                        train_ratio<1 to use a random subset for meta-training
                        (default: 1)

Running mesa-example.ipynb

We include a highly imbalanced dataset Mammography (#majority class instances = 10,923, #minority class instances = 260, imbalance ratio = 42.012) and its variants with flip label noise for quick testing and visualization of MESA and other baselines. You can use mesa-example.ipynb to quickly:

Please check mesa-example.ipynb for more details.

Visualization and Results

From mesa-example.ipynb

Class distribution of Mammography dataset

image

Visualize the meta-training process

<!-- ![image](https://github.com/ZhiningLiu1998/figures/blob/master/mesa/meta-training-process.png) --> <p align="center"> <img src="https://github.com/ZhiningLiu1998/figures/blob/master/mesa/meta-training-process.png" /> </p>

Comparison with baseline methods

image

Other results

Dataset description

image

Comparisons of MESA with under-sampling-based EIL methods

image

Comparisons of MESA with over-sampling-based EIL methods

image

Comparisons of MESA with resampling-based EIL methods

image

Miscellaneous

Check out our previous work Self-paced Ensemble (ICDE 2020).
It is a simple heuristic-based method, but being very fast and works reasonably well.

This repository contains:

NOTE: The implementations of the above baseline methods are based on imbalanced-algorithms and imbalanced-learn.

References

#Reference
[1]N. V. Chawla, A. Lazarevic, L. O. Hall, and K. W. Bowyer, Smoteboost: Improving prediction of the minority class in boosting. in European conference on principles of data mining and knowledge discovery. Springer, 2003, pp. 107–119
[2]S. Wang and X. Yao, Diversity analysis on imbalanced data sets by using ensemble models. in 2009 IEEE Symposium on Computational Intelligence and Data Mining. IEEE, 2009, pp. 324–331.
[3]Sheng Chen, Haibo He, and Edwardo A Garcia. 2010. RAMOBoost: ranked minority oversampling in boosting. IEEE Transactions on Neural Networks 21, 10 (2010), 1624–1642.
[4]C. Seiffert, T. M. Khoshgoftaar, J. Van Hulse, and A. Napolitano, Rusboost: A hybrid approach to alleviating class imbalance. IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and Humans, vol. 40, no. 1, pp. 185–197, 2010.
[5]R. Barandela, R. M. Valdovinos, and J. S. Sanchez, New applications´ of ensembles of classifiers. Pattern Analysis & Applications, vol. 6, no. 3, pp. 245–256, 2003.
[6]X.-Y. Liu, J. Wu, and Z.-H. Zhou, Exploratory undersampling for class-imbalance learning. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 39, no. 2, pp. 539–550, 2009.
[7]Zhining Liu, Wei Cao, Zhifeng Gao, Jiang Bian, Hechang Chen, Yi Chang, and Tie-Yan Liu. 2019. Self-paced Ensemble for Highly Imbalanced Massive Data Classification. 2020 IEEE 36th International Conference on Data Engineering (ICDE). IEEE, 2020, pp. 841-852.
[8]Guillaume Lemaître, Fernando Nogueira, and Christos K. Aridas. Imbalanced-learn: A python toolbox to tackle the curse of imbalanced datasets in machine learning. Journal of Machine Learning Research, 18(17):1–5, 2017.

Contributors ✨

Thanks goes to these wonderful people (emoji key):

<!-- ALL-CONTRIBUTORS-LIST:START - Do not remove or modify this section --> <!-- prettier-ignore-start --> <!-- markdownlint-disable --> <table> <tr> <td align="center"><a href="http://zhiningliu.com"><img src="https://avatars.githubusercontent.com/u/26108487?v=4?s=100" width="100px;" alt=""/><br /><sub><b>Zhining Liu</b></sub></a><br /><a href="#ideas-ZhiningLiu1998" title="Ideas, Planning, & Feedback">🤔</a> <a href="https://github.com/ZhiningLiu1998/mesa/commits?author=ZhiningLiu1998" title="Code">💻</a></td> </tr> </table> <!-- markdownlint-restore --> <!-- prettier-ignore-end --> <!-- ALL-CONTRIBUTORS-LIST:END -->

This project follows the all-contributors specification. Contributions of any kind welcome!