tanogan

TAnoGAN

Citekey	BasharNayak2020TAnoGAN
Source Code	https://github.com/mdabashar/TAnoGAN
Learning type	semi-supervised
Input dimensionality	multivariate

Dependencies

• python 3
• pytorch
• scikit-learn

Notes

TAnoGan outputs anomaly scores for windows with strides. The results require post-processing. The scores for each point can be assigned by aggregating the anomaly scores for each window the point is included in.

You can use the following code snippet for the post-processing step in TimeEval (default parameters directly filled in from the source code):

from timeeval.utils.window import ReverseWindowing
# post-processing for TAnoGan
def post_tanogan(scores: np.ndarray, args: dict) -> np.ndarray:
    length = args.get("dataset_details").length  # type: ignore
    window_size = args.get("hyper_params", {}).get("window_size", 30)
    scores = np.repeat(scores, repeats=window_size)
    result = np.full(shape=length, fill_value=np.nan)
    result[:scores.shape[0]] = scores
    return result