GAE_Unsupervised-Outlier-detection

我们的GAE项目共包含图生成模块、图自编码器模块。使用方法如下：

• 1.首先将ODDS公开数据集进行标准化处理；
• 2.将处理后的数据集转变成.txt文件格式，并输入图生成函数，依据数据集的不同，调整参数。
• 3.将标准化后的数据集X与生成的图共同输入GAE函数进行训练，训练结果以及参数都在GAE函数中有详细说明。

Our GAE project contains a total of graph generation module and graph autoencoder module. The usage methods are as follows.

• 1. first normalize the ODDS public dataset.
• 2. Convert the processed dataset into .txt file format and input the graph construct function, adjusting the parameters according to the dataset.
• 3. Input the normalized dataset X together with the generated graph into the GAE function for training, and the training results and parameters are detailed in the GAE function.

https://github.com/yzhao062/anomaly-detection-resources Toolbox of outlier detection

https://elki-project.github.io/ Toolbox of outlier detection