outlier-hub 0.0.13

Outlierhub, a collection of machine learning datasets.

OutlierHub

a curated hub for outlier and anomaly datasets built on top of datastack.

Currentyly supported datasets:

• ATIS
• ARRHYTHMIA
• KDD
• REUTERS
• TREC
• MNIST
• Fashion-MNIST
• EMNIST
• Newsgroups

Furthermore, OutlierHub supports the following toy datasets:

• Half Moons
• Nested Circles
• Noisy x -> x^3 regression
• Uniform Noise
• Gaussian Cluster
• Circular Segement
• XOR Squares

Install

pip install outlier-hub

For the latest version, clone or download the repository and cd into OutlierHub's root folder and run

pip install src/

Usage

Each dataset has a factory that can be instantiated with a StorageConnector providing the IO operations. After factory instantiation, the factory is able to create a dataset iterator when calling its member method get_dataset_iterator(config={...})

Copyright 2020 Fraunhofer IAIS

For license please see: https://github.com/fraunhofer-iais/outlierhub/blob/master/LICENSE