p60975123
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

!9874 Add checking the threshold of anomaly detection 

From: @zhangxinfeng3
Reviewed-by: @zichun_ye,@wang_zi_dong
Signed-off-by: @zichun_ye
This commit is contained in:
mindspore-ci-bot 2020-12-14 13:55:37 +08:00 committed by Gitee
parent 6a5cf1c711 2f2a10b8a3
commit 63fcab9d86
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
mindspore/nn/probability/toolbox/anomaly_detection.py
View File

@ -15,6 +15,7 @@
"""Toolbox for anomaly detection by using VAE."""
import numpy as np
from mindspore._checkparam import Validator
from ..dpn import VAE
from ..infer import ELBO, SVI
from ...optim import Adam
@ -26,7 +27,7 @@ class VAEAnomalyDetection:
Toolbox for anomaly detection by using VAE.
Variational Auto-Encoder(VAE) can be used for Unsupervised Anomaly Detection. The anomaly score is the error
between the X and the reconstruction. If the score is high, the X is mostly outlier.
between the X and the reconstruction of X. If the score is high, the X is mostly outlier.
Args:
encoder(Cell): The Deep Neural Network (DNN) model defined as encoder.
@ -84,6 +85,7 @@ class VAEAnomalyDetection:
Returns:
Bool, whether the sample is an outlier.
"""
threshold = Validator.check_positive_float(threshold)
score = self.predict_outlier_score(sample_x)
return score >= threshold