Net campus2015 antimomusone

3. 2015 Avanade Inc. All Rights Reserved. Agenda 2 Machine Learning Overview 1 Avanade Italy 3 Azure Machine Learnig 4 Business Case | 3 Copyright 2014 Avanade Inc. All Rights Reserved. The Avanade name and logo are registered trademarks in the US and other countries.

4. 4 Chi Avanade Italy Avanade, nata nel 2000 dalla joint venture tra Microsoft e Accenture, il principale Technology Integrator a livello mondiale specializzato nello sviluppo e nellimplementazione di soluzioni su tecnologia Microsoft per le grandi aziende. Capacit imprenditoriale Soluzioni per qualsiasi settore Acceleratore di soluzioni trasversali Specializzati su piattaforma Microsoft Architetture e asset Risultati realizzati Piattaforme Enterprise Prodotti allavanguardia Impegno per la ricerca e sviluppo

5. Avanade Italy: I nostri numeri CAGLIARI Nata nel Settembre 2000 (tra i Paesi fondatori) 6 Uffici 700 dipendenti Fatturato 80 milioni di Euro ROMA

6. What is Machine Learning ?

7. Machine Learning / Predictive Analytics Vision Analytics Recommenda-tion engines Advertising analysis Weather forecasting for business planning Social network analysis Legal discovery and document archiving Pricing analysis Fraud detection Churn analysis Equipment monitoring Location-based tracking and services Personalized Insurance Machine learning & predictive analytics are core capabilities that are needed throughout your business

8. Machine Learning Overview Formal definition: The field of machine learning is concerned with the question of how to construct computer programs that automatically improve with experience - Tom M. Mitchell Another definition: The goal of machine learning is to program computers to use example data or past experience to solve a given problem. Introduction to Machine Learning, 2nd Edition, MIT Press ML often involves two primary techniques: Supervised Learning: Finding the mapping between inputs and outputs using correct values to train a model Unsupervised Learning: Finding patterns in the input data (similar to Density Estimates in Statistics)

9. Machine Learning Data: A B C D E F G H I J K L M N O P Q R S T U V W X Y Z Rules, or Algorithms: about, Learning, language Spelling and sounding builds words Learning about language. Words build sentences Learning, or Abstraction: Any new understanding proceeds from previous knowledge.

10. Traditional programming VS Machine Learning Computer Data Program Output Computer Data Output Program Traditional Programming Machine Learning

11. No, more like gardening Gardener = You Seeds = Algorithms Nutrients = Data Plants = Programs

12. Sample Application Web search Computational biology Finance E-commerce Space exploration Robotics Information extraction Social networks Debugging [Your favorite area]

13. Types of Learning Supervised (inductive) learning Training data includes desired outputs Unsupervised learning Training data does not include desired outputs Semi-supervised learning Training data includes a few desired outputs

14. Machine Learning Problem Classification or Categorization Clustering Regression Dimensionality reduction Supervised Learning Unsupervised Learning DiscreteContinuous

15. Machine Learning Example Predict function F(X) for new examples X Discrete F(X): Classification Continuous F(X): Regression F(X) = Probability(X): Probability estimation Given examples of a function (X, F(X))

16. Machine Learning Example Training: given a training set of labeled examples {(x1,y1), , (xN,yN)}, estimate the prediction function f by minimizing the prediction error on the training set Testing: apply f to a never before seen test example x and output the predicted value y = f(x) output prediction function Image feature y = f(x) Apply a prediction function to a feature representation of the image to get the desired output: F( ) = apple F( ) = tomato F( ) = dog

17. Supervised Learning 1. Used when you want to predict unknown answers from answers you already have 2. Data is divided into two parts: the data you will use to teach the system (data set), and the data to test the algorithm (test set) 3. After you select and clean the data, you select data points that show the right relationships in the data. The answers are labels, the categories/columns/attributes are features and the values arevalues. 4. Then you select an algorithm to compute the outcome. (Often you choose more than one) 5. You run the program on the data set, and check to see if you got the right answer from the test set. 6. Once you perform the experiment, you select the best model. This is the final output the model is then used against more data to get the answers you need

18. Supervised Learning 1. Car 2. Not Car

19. Unsupervised Learning 1. Used when you want to find unknown answers mostly groupings - directly from data 2. No simple way to evaluate accuracy of what you learn 3. Evaluates more vectors, groups into sets or classifications 4. Start with the data 5. Apply algorithm 6. Evaluate groups

20. Unsupervised Learning Example 1 example A Example 2 example B Example 3 example C example A example B example C Example 1 Example 2 Example 3 The clustering strategies have more tendency to transitively group points even if they are not nearby in feature space

21. Azure Learning Machine Workflow Data Its all about the data. Heres where you will acquire, compile, and analyze testing and training data sets for use in creating Azure Machine Learning predictive models. Create the model Use various machine learning algorithms to create new models that are capable of making predictions based on inferences about the data sets. Evaluate the model Examine the accuracy of new predictive models based on ability to predict the correct outcome, when both the input and output values are known in advance. Accuracy is measured in terms of confidence factor approaching the whole number one. Refine and evaluate the model Compare, contrast, and combine alternate predictive models to find the right combination(s) that can consistently produce the most accurate results. Deploy the model Expose the new predictive model as a scalable cloud web service, one that is easily accessible over the Internet by any web browser or mobile client. Test and use the model Implement the new predictive model web service in a test or production application scenario.

22. Azure Machine Learning algorithms Classification algorithms These are used to classify data into different categories that can then be used to predict one or more discrete variables, based on the other attributes in the dataset. Regression algorithms These are used to predict one or more continuous variables, such as profit or loss, based on other attributes in the dataset. Clustering algorithms These determine natural groupings and patterns in datasets and are used to predict grouping classifications for a given variable. Supervised learning Classification algorithms Regression algorithms Unsupervised learning Clustering algorithms

23. Steps Recap

24. Deploying a prediction model The deployment of a new prediction model takes the form of exposing a web service which can then be invoked via the Representational State Transfer (REST) protocol. Azure Machine Learning web services can be called via two different exposed interfaces: Single, request/response-style calls. Batch style calls, where multiple input records are passed into the web service in a single call and the corresponding response contains an output list of predictions for each input record. When a new machine learning prediction model is exposed on the Web, it performs the following operations: New input data is passed into the web service in the form of a JavaScript Object Notation (JSON) payload. The web service then passes the incoming data as inputs into the Azure Machine Learning prediction model engine. The Azure Machine Learning model then generates a new prediction based on the input data and returns the new prediction results to the caller via a JSON payload.

25. AzureML Why AzureML? Setting up a Microsoft Azure Account Setting up a Storage Account Loading Data Setting up an AzureML Workspace Accessing AzureML Studio AzureML Studio Tour

26. Azure ML Getting Started Option 1 Take advantage of a free Azure trial offer at http://azure.microsoft.com/en-us/pricing/free-trial Option 2 Take advantage of the (free) Azure Machine Learning trial offer at https://studio.azureml.net/Home How to get started ?

27. Azure ML Getting Started Option 1 Take advantage of a free Azure trial offer at http://azure.microsoft.com/en-us/pricing/free-trial Option 2 Take advantage of the (free) Azure Machine Learning trial offer at https://studio.azureml.net/Home

28. Azure Machine Learning workspaces Experiments Azure ML Studio Web Services Datasets Modules

29. Demo Azure Machine Learning Portal Overview

30. Create your first Azure Machine Learning experiment Defining the problem you want to solve e.g. figure out if you like certain movie whats another movie you should watch? (movie recommendation ) Designing a Solution Using AzureML 1. Creating an experiment 2. Load a Data Set 3. Create the Experiment 4. Add Transformations and Filters 5. Create the Experiment Path and apply Algorithms 6. Save and Run the Model 7. Publish the Model 8. Use the Model Saving and Running Publishing and Accessing

31. Demo Azure Machine Learning Demo Predict ratings for a given user and movie Recommend movie to a given user Find users related to a given user Find items related to a given item

32. API examples Difference in Proportions Test Lexicon Based Sentiment Analysis Forecasting-Exponential Smoothing Forecasting - ETS+STL Forecasting-AutoRegressive Integrated Moving Average (ARIMA) Normal Distribution Quantile Calculator Normal Distribution Probability Calculator Normal Distribution Generator Binomial Distribution Probability Calculator Binomial Distribution Quantile Calculator Binomial Distribution Generator Multivariate Linear Regression Survival Analysis Binary Classifier Cluster Model datamarket.azure.com

33. The rest is up to you

34. Sell your Azure Machine Learnig Service Publish Azure Machine Learning Web Service to the Azure Marketplace http://azure.microsoft.com/en-us/documentation/articles/machine-learning-publish-web-service-to- azure-marketplace http://datamarket.azure.com Azure Marketplace

35. Check out the Machine Learning marketplace at datamarket.azure.com Learn how to on-board to the marketplace off azure.com at Machine Learning Center Visit the Microsoft booth to talk to the team Next Steps

36. Avanade Business Case Connected Mine Walk Through ( IOT + Azure Machine Learning)

38. 2015 Avanade Inc. All Rights Reserved. Follow us on: http://www.avanade.com/it-it/home http://blog.avanade.com/inside-avanade/ https://it-it.facebook.com/avanade.italy https://twitter.com/avanadeitaly https://www.linkedin.com/company/avanade-italy?trk=biz-brand-tree- co-name

Net campus2015 antimomusone

Documents

machine learning data

field of machine learning

goal of machine learning

types of learning

sentences learning

model unsupervised learning

chi avanade italy avanade

avanade dal