Autogluon

AutoGluon: AutoML for Text, Image, and Tabular Data
Under Apache License 2.0
By awslabs

deep-learning machine-learning pytorch object-detection computer-vision natural-language-processing data-science scikit-learn image-classification automl automated-machine-learning gluon mxnet transfer-learning tabular-data ensemble-learning neural-architecture-search structured-data hyperparameter-optimization autogluon

AutoML for Text, Image, and Tabular Data

AutoGluon automates machine learning tasks enabling you to easily achieve strong predictive performance in your applications. With just a few lines of code, you can train and deploy high-accuracy machine learning and deep learning models on text, image, and tabular data.

Example

```python

First install package from terminal:
python3 -m pip install -U pip
python3 -m pip install -U setuptools wheel
python3 -m pip install -U "mxnet<2.0.0"
python3 -m pip install autogluon # autogluon==0.3.1

from autogluon.tabular import TabularDataset, TabularPredictor
train_data = TabularDataset('https://autogluon.s3.amazonaws.com/datasets/Inc/train.csv')
test_data = TabularDataset('https://autogluon.s3.amazonaws.com/datasets/Inc/test.csv')
predictor = TabularPredictor(label='class').fit(train_data, time_limit=120) # Fit models for 120s
leaderboard = predictor.leaderboard(test_data)
```

News

Announcement for previous users: The AutoGluon codebase has been modularized into namespace packages, which means you now only need those dependencies relevant to your prediction task of interest! For example, you can now work with tabular data without having to install dependencies required for AutoGluon's computer vision tasks (and vice versa). Unfortunately this improvement required a minor API change (eg. instead of from autogluon import TabularPrediction, you should now do: from autogluon.tabular import TabularPredictor), for all versions newer than v0.0.15. Documentation/tutorials under the old API may still be viewed for version 0.0.15 which is the last released version under the old API.

Resources

See the AutoGluon Website for documentation and instructions on:
- Installing AutoGluon
- Learning with tabular data
- Tips to maximize accuracy (if benchmarking, make sure to run fit() with argument presets='best_quality').

Learning with text data

Learning with image data

More advanced topics such as Neural Architecture Search

Scientific Publications

AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data (Arxiv, 2020)

Fast, Accurate, and Simple Models for Tabular Data via Augmented Distillation (NeurIPS, 2020)

Multimodal AutoML on Structured Tables with Text Fields (ICML AutoML Workshop, 2021)

Articles

AutoGluon for tabular data: 3 lines of code to achieve top 1% in Kaggle competitions (AWS Open Source Blog, Mar 2020)

Accurate image classification in 3 lines of code with AutoGluon (Medium, Feb 2020)

AutoGluon overview & example applications (Towards Data Science, Dec 2019)

Hands-on Tutorials

From HPO to NAS: Automated Deep Learning (CVPR 2020)

Practical Automated Machine Learning with Tabular, Text, and Image Data (KDD 2020)

Train/Deploy AutoGluon in the Cloud

AutoGluon-Tabular on AWS Marketplace

Running AutoGluon-Tabular on Amazon SageMaker

Running AutoGluon Image Classification on Amazon SageMaker

Citing AutoGluon

If you use AutoGluon in a scientific publication, please cite the following paper:

Erickson, Nick, et al. "AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data." arXiv preprint arXiv:2003.06505 (2020).

BibTeX entry:

bibtex @article{agtabular, title={AutoGluon-Tabular: Robust and Accurate AutoML for Structured Data}, author={Erickson, Nick and Mueller, Jonas and Shirkov, Alexander and Zhang, Hang and Larroy, Pedro and Li, Mu and Smola, Alexander}, journal={arXiv preprint arXiv:2003.06505}, year={2020} }

If you are using AutoGluon Tabular's model distillation functionality, please cite the following paper:

Fakoor, Rasool, et al. "Fast, Accurate, and Simple Models for Tabular Data via Augmented Distillation." Advances in Neural Information Processing Systems 33 (2020).

BibTeX entry:

bibtex @article{agtabulardistill, title={Fast, Accurate, and Simple Models for Tabular Data via Augmented Distillation}, author={Fakoor, Rasool and Mueller, Jonas W and Erickson, Nick and Chaudhari, Pratik and Smola, Alexander J}, journal={Advances in Neural Information Processing Systems}, volume={33}, year={2020} }

If you use AutoGluon's multimodal text+tabular functionality in a scientific publication, please cite the following paper:

Shi, Xingjian, et al. "Multimodal AutoML on Structured Tables with Text Fields." 8th ICML Workshop on Automated Machine Learning (AutoML). 2021.

BibTeX entry:

bibtex @inproceedings{agmultimodaltext, title={Multimodal AutoML on Structured Tables with Text Fields}, author={Shi, Xingjian and Mueller, Jonas and Erickson, Nick and Li, Mu and Smola, Alex}, booktitle={8th ICML Workshop on Automated Machine Learning (AutoML)}, year={2021} }

AutoGluon for Hyperparameter and Neural Architecture Search (HNAS)

AutoGluon also provides state-of-the-art tools for neural hyperparameter and architecture search, such as for example ASHA, Hyperband, Bayesian Optimization and BOHB. To get started, checkout the following resources

General introduction into HNAS

Introduction into HNAS with AutoGluon

Example notebook

Example scripts for efficient multi-fidelity HNAS of PyTorch neural network models

Also have a look at our paper "Model-based Asynchronous Hyperparameter and Neural Architecture Search" arXiv preprint arXiv:2003.10865 (2020).

bibtex @article{abohb, title={Model-based Asynchronous Hyperparameter and Neural Architecture Search}, author={Klein, Aaron and Tiao, Louis and Lienart, Thibaut and Archambeau, Cedric and Seeger, Matthias}, journal={arXiv preprint arXiv:2003.10865}, year={2020} }

AutoGluon for Constrained Hyperparameter Optimization

AutoGluon includes an algorithm for constrained hyperparameter optimization. Check out our paper applying it to optimize model performance under fairness constraints: "Fair Bayesian Optimization", AIES (2021).

bibtex @article{fairbo, title={Fair Bayesian Optimization}, author={Perrone, Valerio and Donini, Michele and Zafar, Bilal Muhammad and Schmucker, Robin and Kenthapadi, Krishnaram and Archambeau, Cédric}, journal={AIES}, year={2021} }

License

This library is licensed under the Apache 2.0 License.

Contributing to AutoGluon

We are actively accepting code contributions to the AutoGluon project. If you are interested in contributing to AutoGluon, please read the Contributing Guide to get started.