v1.1.0

What's Changed

• Add two detectors: GADNR by @YingtongDou and DMGD by @kayzliu
• Add tutorial for score conversion
• Add tutorial for GPU training
• Multiple bugs fixed

Greatly appreciate our community contributors helping improve PyGOD: @OldPanda, @ParthaPratimBanik, @ahmed3amerai

Full Changelog: v1.0.0...v1.1.0

v1.0.0

We are thrilled to release PyGOD v1.0.0, a comprehensive open-source graph outlier detection library in the PyG ecosystem.

PyGOD contains more than 10 latest graph outlier detectors, which are built on PyTorch and PyG. It features:

• unified and simple API: detector.fit, detector.predict
• full documentation and examples at docs.pygod.org
• all you need to prepare a PyG Data object

New features in recent versions:

• built-in data and benchmark: utils.load_data
• modularized components: nn.encoder, nn.decoder, nn.fuctional, etc.
• extensive utility functions: metric, generator, utils, etc.

If you encounter a bug or have any suggestions please fill an issue or reach us via email at dev@pygod.org. Also, feel free to try it out with your code! We appreciate every star, fork, and follow.

v0.4.0

We are excited to announce the final pre-alpha release, PyGOD v0.4, which marks a major milestone in our development. Following bug fixes and minor improvements, we plan to release v1.0. Your feedback and suggestions are appreciated. ⚠️ Please note that this version is NOT forward compatible and some APIs have changed. Here are the major changes in this release:

Enhanced Base Class

• Detector: base class for all detectors.
• DeepDetector: base class for all deep learning based detectors.

Simplied APIs

• Removed predict_proba and predict_confidence.
• Use predict(return_prob=True, return_conf=True) instead.

Modularized Detectors

We now introduce multiple modules to improve the code reusability and extendibility.

• nn: all base models inherit torch.nn.Module
• nn.encoder:
• nn.decoder:
• nn.functional: loss function, etc.
  Also, we changed the name of several modules to improve the clarity.
• models→detector
• metrics→metric

More Utility Functions

• to_edge_score: edge outlier score converter
• to_graph_score: graph outlier score converter
• init_detector: detector initializer
• init_nn: neural network initializer

Updated Requirements

• PyGOD now requires Python 3.8+
• PyTorch 2.0 and PyG 2.3.0 support
• Enabled model compile via detector(compile_model=True) (beta)

And Many More

• More comprehensive test coverage (almost 100%)
• Reorganized documentation for better readability
• Merge MLPAE and GCNAE to GAE
• Most of the deep detectors support specifying various backbone from PyG
• Retrieve learned embedding from fitted deep detectors with save_emb=True by detector.emb

v0.3.1

What's Changed

• add edge drop probability to structural outlier injection
• update benchmark script with more datasets.
• multiple minor fixes by @cshjin @YingtongDou @kayzliu

New Contributors

• @cshjin made their first contribution in #40

v0.3.0

What's New

• We release the first comprehensive node-level graph outlier detection benchmark, and the paper is available on arXiv. See benchmark and data for more details.
• Add new models SCAN, Radar, and ANOMALOUS.
• Accelerate GAAN and CONAD by vectorization. Up to 40x faster.
• Add a new metric eval_ndcg.

v0.2.0

What's New

• Our paper is available on arXiv.
• We enable most of the models to train with minbatch, see model list for supported models. @kayzliu @xyvivian @aha12345678
• Add new models CoLA (beta) and ANEMONE (beta) by @harvardchen
• The first community contributor @zhiming-xu add a new model CONAD.
• Add new metric eval_average_precision by @YingtongDou.
• Improved device setting by @yzhao062

v0.1.1

Many key applications depend on graph data. To tailor this need, we just open-sourced the first comprehensive graph outlier detection library--PyGOD.

PyGOD contains more than 10 latest graph outlier detectors, which are built on PyTorch and PyG. It features:

• unified and simple API as PyOD: using GNNs for outlier detection within 5 lines of code
• full documentation and examples
• for both academic use and industry app, all you need to prepare is the data in PyG format.

PyGOD is a collaborative effort among UIC, CMU, ASU, IIT, and BUAA. We commit to providing long-term maintenance and keep adding new models to the library. It is also our goal to promote graph outlier detection methods to broader audiences. If you encounter a bug or have any suggestions please fill an issue or reach us via email dev@pygod.org. Also, feel free to try it out with your code!

We appreciate every star, fork, and follow.