MICROSOFT AZURE MACHINE LEARNING Oscar Naim Microsoft
Microsoft Azure Machine Learning
What is Machine Learning? Azure Machine Learning: How it works Azure Machine Learning in action Get started Contents
What is Machine Learning? “ Delivering on one of the old dreams Predictive computing of Microsoft co-founder Bill Gates: Computers that can see, hear systems become smarter ” and understand. with experience John Platt Distinguished scientist at Microsoft Research
Why Learn? “ Delivering on one of the old dreams of Microsoft co-founder Bill Gates: Learn it when you can’t code it (e.g. speech recognition) Computers that can see, hear ” and understand. Learn it when you can’t scale it (e.g. recommendations) John Platt Distinguished scientist at Learn it when you have to adapt/personalize Microsoft Research (e.g. predictive typing) Learn it when you can’t track it (e.g. robot control)
The United States Postal Service processed over 150 billion pieces of mail in 2013 — far too much for efficient human sorting. But as recently as 1997, only 10% of hand-addressed mail was successfully sorted automatically.
The challenge in automation is enabling computers to interpret endless variation in handwriting.
By providing feedback, the Postal Service was able to train computers to accurately read human handwriting. T oday, with the help of machine learning, over 98% of all mail is successfully processed by machines.
Microsoft & Machine Learning 15 years of realizing innovation 1999 2004 2005 2008 2010 2012 2014 Computers Microsoft SQL Server Bing Maps Microsoft Successful, Microsoft work on users search engine enables ships with ML Kinect can real-time, launches behalf, filtering built with data mining traffic- watch users speech-to- Azure Machine junk email machine prediction gestures speech Learning learning service translation “ Machine learning is pervasive throughout John Platt, ” Microsoft products. Distinguished scientist at Microsoft Research
Huge set-up costs of tools, expertise, and Expensive compute/storage capacity create unnecessary barriers to entry Siloed Siloed and cumbersome data management Break away data restricts access to data from industry Complex and fragmented tools limit Fragmented limitations participation in exploring data and tools building models Many models never achieve business Deployment value due to difficulties with deploying complexity to production
Azure Machine Learning How it works “ Azure Machine Learning offers a data science experience that is Enable custom predictive directly accessible to business analytics solutions at the analysts and domain experts, reducing complexity and speed of the market broadening participation through ” better tooling. Hans Kristiansen Capgemini
The Environments The Team Azure Portal Azure Ops Team ML Studio Data Scientists ML API service Developers
Web Apps Mobile Apps PowerBI/Dashboards ML API service Developer Azure Portal Azure Portal & ML Studio HDInsight ML API service Azure Storage Azure Ops Team Azure Ops Team Data Scientist Desktop Data
Business users easily access results: from anywhere, on any device Web Apps Mobile Apps PowerBI/Dashboards ML API service ML API service and the Developer Developer • Tested models available as an url that can be called from any end point Azure Portal Azure Portal & ML Studio Azure Portal & ML API service ML Studio HDInsight and the Azure Ops Team and the Data Scientist ML API service • Create ML Studio workspace • Access and prepare data Azure Storage • Assign storage account(s) • Create, test and train models • Monitor ML consumption • Collaborate • See alerts when model is ready • One click to stage for Azure Ops Team Azure Ops Team Data Scientist • Deploy models to web service production via the API service Desktop Data
Fully Easy to use T ested Deploy in managed solutions minutes No software to install, Simple drag, drop and Access to sample Tooled for quick no hardware to manage, connect interface you experiments, tested deployment, hand-off and one portal to view can access and share algorithms, support for and updates and update from anywhere custom R, and over 350 R packages
Azure Machine Learning in action “ There was zero percent chance we were going to take a step backwards and consider a machine learning Real world examples solution that wasn’t well -established ” and proven effective in the cloud. Kristian Kimbro Rickard MAX451
Smart Buildings The Center for Building Performance and Diagnostics uses weather forecasts, real-time temperature reads, and behavioral research data to optimize building heating and cooling “ The ease of implementation systems in real-time. makes machine learning Key Benefits accessible to a larger User friendly set up and integration with number of investigators with • existing systems various backgrounds — even Seamless data handling ” • non-data scientists. Accessible and easy to use across • backgrounds Bertrand Lasternas Quickly compare algorithms • Carnegie Mellon
Demand Forecasting Pier 1 partnered with MAX451 to delight loyalty customers by using historical and “ behavioral data to predict what products We are especially pleased they want next. that our analysts can focus on the results and not worry about the complex Key Benefits ” algorithms behind the scenes. Ease of use across skillsets • Fast time to meaningful results • Accessible via the cloud • Andrew Laudato Pier 1 Imports
Investment Optimization Icertis, a cloud solutions provider, built a predictive model using past performance data “ to determine the optimal locations for its The standout benefit for us clients to build new retail stores. was to quickly build and test predictive models and verify their results. There is no Key Benefits cognitive overhead to learn Quickly build, test and verify models new scripting or coding • No new scripting or coding languages ” language. • Easily import and modify algorithms • developed outside the solution Yogesh Dandawate Icertis Applied Cloud
Image Churn Ad detection & analysis targeting classification Imagine what machine Equipment learning could do for Recommendations Forecasting monitoring your business. Spam Fraud Anomaly filtering detection detection
“ Learn more and Microsoft has a solid track record for creating user-friendly tools, sign up for a and Pier 1 is helping prove free trial of Azure Microsoft can take something as complex as machine learning and make it accessible via the azure.com/ml ” cloud Andy Laudato Pier 1 Imports
Recommend
More recommend
