normalizing_flows

Normalizing Flows

Citekey	ryzhikov2019normalizing
Source code	own
Learning type	supervised
Input dimensionality	multivariate

Dependencies

• python 3
• pytorch

Notes

Normalizing Flows outputs anomaly scores for windows. 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 Normalizing Flows
def post_nf(scores: np.ndarray, args: dict) -> np.ndarray:
    window_size = args.get("hyper_params", {}).get("window_size", 20)
    return ReverseWindowing(window_size=window_size).fit_transform(scores)