https://store-images.s-microsoft.com/image/apps.37305.ee5d4d3e-2a1a-48da-8831-643830a3c1f1.d6838687-9460-4365-a308-10243726009a.0cfe02d0-2d92-4fa2-bfb7-0fb744ab33e9

Unsupervised Anomaly Detection Module

BRFRame Inc

Unsupervised Anomaly Detection Module

BRFRame Inc

Anomaly Detection IoT Edge Module using Unsupervised Model (with Python, CNTK)

Generally, there needs labeled data for the abnormal section to detect anomalies in the dataset when using supervised learning model so in the past to define abnormal section in the history data, we should match and find it with fault-check log or failure data and these kinds of work would take a lot of time and sometimes are not accurate.

In order to solve these kinds of problem, the section clustering results using unsupervised learning pattern detection based on the various sensor data are provided with a model that automatically categorizes abnormal sections.

So this can be used in prediction model as a method of Auto-Labeling

To see more details and some guidelines, please click Edge Module Structure and Guideline.


Minimum hardware requirements :
Linux(x64), Minimum 8GB RAM, 17G Storage, Python 3.5, Iot Edge Runtime

https://store-images.s-microsoft.com/image/apps.5862.ee5d4d3e-2a1a-48da-8831-643830a3c1f1.d6838687-9460-4365-a308-10243726009a.aaed3c68-42f6-41ed-849f-2c16424e4ee1
https://store-images.s-microsoft.com/image/apps.5862.ee5d4d3e-2a1a-48da-8831-643830a3c1f1.d6838687-9460-4365-a308-10243726009a.aaed3c68-42f6-41ed-849f-2c16424e4ee1