NI偷偷發布了分析和機器學習工具包

02-04

LabVIEW 2017 Analytics and Machine Learning Toolkit

NI在機器學習的大潮下也發布了機器學習工具包

模型訓練支持Windows平台，部署支持Windows和NI Linux RT平台，演算法用的是之前故障預診斷的 IMS Center Watchdog Agent工具包，不過沒有特徵提取的部分，只有特徵處理模型訓練和部署的功能

Analytics and

Machine Learning Toolkit

July 2017,

377059A-01

The LabVIEW

Analytics and Machine Learning toolkit includes VIs for training machine

learning models that discover patterns in large amounts of data. You can deploy

trained machine learning models to recognize patterns in new data.

The following figure demonstrates the machine learning process when you use the Analytics and Machine Learning Toolkit.

The previous machine learning process contains the following components:

Data

Collection—Collects data from data acquisition devices or other sources.

Feature Extraction

and Reduction—Creates features based on your domain knowledge of data and

reduces the dimension of data so that you can apply machine learning algorithms

to the training data.

Model

Creation—Trains models using machine learning algorithms and training data.

Model

Validation—Validates models using model evaluation metrics and test data.

Deployment—Deploys trained models.

The Analytics and Machine Learning Toolkit uses algorithms from the IMS Center Watchdog Agent.

基本上最常用的幾個機器學習的演算法都有了，感覺比之前NI Lab裡面的演算法完善了不少，把底層的細節封裝掉了，並且模型的訓練和部署分離，以後可以考慮實現終端節點模型的動態更新。