Ta spletna stran hrani piškotke, da bi vam zagotovili boljšo uporabniško izkušnjo in popolno funkcionalnost te strani.

Analitične piškotke uporabljamo s storitvijo Google Analytics, samo z vašo privolitvijo. Sprejemam Zavrnitev Več informacij

Model Interpretability and explainability with Azure Machine Learning

Explaining Machine Learning models have become black box solutions. In the past two years, many tools and python packages have been developed to deliver more insights into understanding and explaining the outcome of the ML model and why it behaved the way it has. Addressing the questions like: how sure can we be, if the prediction is correct, what information does the accuracy matrix hold, how different would have the outcome been, had it have different input parameters set. Introducing also the principles of ML, such as fairness of models, accountability, and transparency. Demos will include Azure Machine Learning service and different Python packages to address topics of explainability and interpretability of models.

Tomaž Kaštrun

/

Tomaž Kaštrun is a SQL Server developer and data scientist with more than 15 years of experience in business warehousing, development, ETL, database administration, and query tuning. He holds over 15 years of experience in data analysis, data mining, statistical research, and machine learning. He is a Microsoft SQL Server MVP for data platform and has been working with Microsoft SQL Server since version 2000. He is a blogger, author of many articles, a frequent speaker at the community and Microsoft events. He is an avid coffee drinker who is passionate about fixed-gear bikes. In 2018 he co-authored the book "SQL Server 2017 Machine Learning Services with R".