4.1. Pre-installed AITsΒΆ

Qunomon includes pre-installed AITs that are ready to use. While the quality metrics that can be measured with AITs are not limited to these, reviewing their configurations and implementations may serve as a reference for developing new AITs. The repository is located at https://github.com/aistairc/{name_of_AIT}. Please refer to Section 2. Operating Conditions and Installation for details on downloading methods and directory structure.

  • alyz_dataset_table_counts_comb_two_attr

    • Specifies unnecessary (impossible) attribute value combinations in tabular data, and calculates the count and proportion of these unwanted data present in the table.

    • AIT users can use this summary information to understand the unhealthiness of attribute values based on their occurrence trends.

  • eval_dataset_image_diversity_vae

    • Calculates the feature metrics of evaluation image data using a VAE model trained on features of training image data.

    • The closer these feature differences are to zero, the more comprehensively the evaluation image data is considered to encompass the features of the training data.