ARTIFICIAL INTELLIGENCE (AI): NEW FACE OF IT Sandesh Sudhakaran K S Sandesh.ks@emc.com Knowledge Sharing Article 2018 Dell Inc. or its subsidiaries.
Table of Contents Purpose... 3 Introduction... 3 What is Artificial Intelligence?... 3 Overview - Microsoft AI platform... 4 What is Machine Learning?... 5 How Machine Learning Works... 5 What is Azure Machine Learning Studio?... 7 Creating Predictive Models using ML Studio...12 Language Understanding Intelligence Service...15 What is a LUIS app?...15 The Microsoft Bot Framework...17 Cognitive Toolkit...18 Conclusion...19 References...19 Disclaimer: The views, processes or methodologies published in this article are those of the author. They do not necessarily reflect Dell EMC s views, processes or methodologies. 2018 Dell EMC Proven Professional Knowledge Sharing 2
Purpose This whitepaper provides an overview of Artificial Intelligence (AI) and explains how it can be used to build smart apps that help organizations be more efficient and enrich people s lives. This paper focuses on the following aspects: Discover how machine learning can be used to build predictive models for AI. Find out how to build intelligent bots that enable conversational communication between humans and AI systems. The practical elements of this paper are based on Microsoft Azure. Introduction This whitepaper provides an introduction to AI, and the ways in which AI technology can be used to achieve more and enrich lives. We'll explore some of the machine learning principles that provide the foundation for AI and discover some of the fundamental techniques that are used to integrate AI capabilities into applications. Large amounts of data, faster processing power, and increasingly smarter algorithms are powering AI applications and associated use cases across consumer, finance, healthcare, manufacturing, transportation & logistics, and government sectors around the world, enabling smarter and more intelligent applications to speak, listen, and make decisions in unprecedented ways. As AI technologies and deployments sweep through virtually every industry, a wide range of use cases are beginning to illustrate the potential business opportunities, and inspire changes to existing business processes leading to newer business models. What is Artificial Intelligence? When we talk about artificial intelligence, it's easy to imagine some dystopian science fiction future where robots have taken over the world and enslaved us. AI is a way to enable people to accomplish more by collaborating with smart software. Think of it as putting a more human face on technology technology that can learn from the vast amounts of data available in the modern world and technology that can understand our kind of language and respond in kind or technology that can see and interpret the world the way that we do. 2018 Dell EMC Proven Professional Knowledge Sharing 3
As our world becomes increasingly digitized, so does our ability to connect with it. Imagine if you could search your surroundings the same way you search the web. Using existing cameras and advances in AI, we can now find things and people in the real world, in real time and take action to improve safety and well-being. Overview - Microsoft AI platform The Microsoft AI platform offers a comprehensive set of flexible AI Services, enterprise-grade AI Infrastructure and modern AI Tools for developers and data scientists to create applications of the future. AI platform consists of 3 core areas: AI Services: Developers can rapidly consume high-level finished services that accelerate development of AI solutions. Compose intelligent applications, customized to your organization s availability, security, and compliance requirements. This collection of powerful APIs enables your apps to intelligently interpret the world and to naturally engage your users through vision, speech, language, and knowledge. With a few lines of code, you can easily build apps that will learn, adapt, and advance over time, across a multitude of devices and platforms. AI Infrastructure: Services and tools backed by a best-of-breed infrastructure with enterprise grade security, availability, compliance, and manageability. Harness the power of infinite scale infrastructure and integrated AI services. Engages customers more naturally and where they already are. The Microsoft Bot Framework makes it easy for anyone to create new experiences and reach at scale. Easily build and deploy across channels including Facebook Messenger, Cortana, Slack, Skype, and now Bing. AI Tools: Leverage a set of comprehensive tools and frameworks to build, deploy, and operationalize AI products and services at scale. Use the extensive set of supported tools and IDEs of your choice and harness the intelligence with massive datasets through deep learning frameworks of your choice. Microsoft s free, easy-to-use, open-source, commercial-grade toolkit trains deep-learning algorithms to learn like the human brain. Adopting programming languages and algorithms you already use, the Microsoft Cognitive Toolkit allows harnessing massive dataset intelligence with uncompromised scaling, speed, accuracy, and compatibility. 2018 Dell EMC Proven Professional Knowledge Sharing 4
A comprehensive set of flexible AI services for any scenario and enterprise-grade AI infrastructure runs AI workloads anywhere at scale. Modern AI tools designed for developers and data scientists helps create AI solutions easily and with maximum productivity. What is Machine Learning? Machine learning is a data analytics technique that teaches computers to do what comes naturally to humans and animals: learn from experience. Machine learning provides the foundation for artificial intelligence. Machine learning algorithms use computational methods to learn information directly from data without relying on a predetermined equation as a model. The model learns from the training cases and then we can use the trained model to make predictions for new data cases. The key to this is to understand that fundamentally, computers are very good at one thing; performing calculations. To have a computer make intelligent predictions from the data, we just need a way to train it to perform the correct calculations. The algorithms adaptively improve their performance as the number of samples available for learning increases. How Machine Learning Works Machine learning uses two types of techniques: Supervised Learning: This trains a model on known input and output data so that it can predict future outputs. Unsupervised Learning: This helps find hidden patterns or intrinsic structures in input data. 2018 Dell EMC Proven Professional Knowledge Sharing 5
Supervised Learning Supervised machine learning builds a model that makes predictions based on evidence in the presence of uncertainty. A supervised learning algorithm takes a known set of input data and known responses to the data (output) and trains a model to generate reasonable predictions for the response to new data. Use supervised learning if you have known data for the output you are trying to predict. Supervised learning uses Classification and Regression techniques to develop predictive models. 1. Classification Techniques predict discrete responses for example, whether an email is genuine or spam, or whether a tumor is cancerous or benign. Classification models classify input data into categories. Typical applications include medical imaging, speech recognition, and credit scoring. Use classification if your data can be tagged, categorized, or separated into specific groups or classes. For example, applications for hand-writing recognition use classification to recognize letters and numbers. In image processing and computer vision, unsupervised pattern recognition techniques are used for object detection and image segmentation. 2. Regression Techniques predict continuous responses for example, changes in temperature or fluctuations in power demand. Typical applications include electricity load forecasting and algorithmic trading. Regression techniques are used when you are working with a data range or if the nature of your response is a real number, such as temperature or the time until failure for a piece of equipment. Unsupervised Learning Unsupervised learning finds hidden patterns or intrinsic structures in data. It is used to draw inferences from datasets consisting of input data without labeled responses. Clustering is the most common unsupervised learning technique. It is used for exploratory data analysis to find hidden patterns or groupings in data. Applications for cluster analysis include gene sequence analysis, market research, and object recognition. For example, if a cell phone company wants to optimize the locations where they build cell phone towers, they can use machine learning to estimate the number of clusters of people relying on their towers. A phone 2018 Dell EMC Proven Professional Knowledge Sharing 6
can only talk to one tower at a time, so the team uses clustering algorithms to design the best placement of cell towers to optimize signal reception for groups, or clusters, of their customers. There are a number of ways we can create a clustering model. One of the most popular clustering techniques is called k-means Clustering. k-means clustering is a partitioning method. The function kmeans partitions data into k mutually exclusive clusters, and returns the index of the cluster to which it has assigned each observation. k-means clustering operates on actual observations and creates a single level of clusters. kmeans treats each observation in your data as an object having a location in space. It finds a partition in which objects within each cluster are as close to each other as possible, and as far from objects in other clusters as possible. You can choose from five different distance measures, depending on the kind of data you are clustering. Each cluster in the partition is defined by its member objects and by its centroid, or center. The centroid for each cluster is the point to which the sum of distances from all objects in that cluster is minimized. kmeans computes cluster centroids differently for each distance measure, to minimize the sum with respect to the measure that you specify. What is Azure Machine Learning Studio? Now that we understand a bit more about what machine learning is, it's time to think about how we can work with data to create predictive models. There are lots of tools, services, and programming languages that data scientists use to prepare a model and test data. We will look at an easy to use service for machine learning and modeling in Microsoft Azure, but the same principles and work flow apply to almost any machine learning framework. Azure machine learning is a cloud service in Azure that enables us to experiment with data and create predictive models. We start with a data set, which contains the data with which we want to experiment, and this data set provides the starting point of an experiment in which we will define a workflow of transformations to apply to the data as we explore it and prepare it for modeling. To access the Azure Machine Learning Studio, we will need a Microsoft account, and the same account is also required to create a Workspace. Once logged in with the Microsoft account we will be able to see the browser-based work bench which is the Azure ML Studio. The ML Studio has 3 sections: 2018 Dell EMC Proven Professional Knowledge Sharing 7
Experiments Web Services Settings The Experiments section is where you will offer predictive analytics and spend the most time. The Web Services section is where we will be able to see the list of web services created. Finally, the Settings section is where we will be able to change the settings for the workspace and share the workspaces created to the team members. The experiment section shows all the experiments in the workspace, with the experiments that are being edited most recently at the top of the list. If any experiment needs to be deleted, select the experiment and click the Delete icon on the bottom of the page. Below is a snapshot from the Microsoft Azure Machine Learning Studio. The workspace comes prepopulated with Samples and when we select the Samples tab, we can see the list of samples experiments available. By clicking on a sample experiment we can see all of its modules. We can clone the experiment and save the editable file of the experiment by clicking on the Save As button at the bottom of the page. 2018 Dell EMC Proven Professional Knowledge Sharing 8
The first set of building a predictive model is uploading a dataset. The New icon at the bottom of the page allows you to do just that. Click on Dataset - From Local File and you can see a dialog where you can upload data in multiple file formats. The New button at the bottom of the page also allows you to create a New Experiment. When we create an experiment, we first need to give our Experiment a name. The left side of the experiment canvas has a palette of modules that are available to build an experiment. We can look through the pre-defined categories to select the modules for building an experiment or search the modules for a specific keyword. 2018 Dell EMC Proven Professional Knowledge Sharing 9
Next is the Web Services section. Clicking on this service brings us to the Dashboard for the Service. The page lists the Parent Experiment that is used to build the service. The page also lists the API access key needed for calling the service. Finally, the page lists both the Request Response and Batch Execution Service. The Configuration page for the Service is where we can go to change the service configuration and description for the Input and Output parameters. 2018 Dell EMC Proven Professional Knowledge Sharing 10
The Settings section is where we can invite other users to our Workspace. To invite co-workers to our Workspace, we click on the Invite More Users icon on the bottom of the page and type in their Microsoft account email address. We can also decide if we need to give the person User or Owner permissions. A User has permission to list, clone and create experiments and datasets in the workspace and the Owner can add, remove and list users with access to the workspace, in addition to what a person with User permission can do. Once the invitation has been sent, we can see the user listed with the role and status in the Settings section. To remove a user from the workspace, we simply need to select the user and click on the Remove icon on the bottom of the page, which will withdraw their access from your workspace. Once a user has been invited to a workspace, they will receive an email with the link to access your workspace. They will need to click on the invitation link in the email to access this workspace. 2018 Dell EMC Proven Professional Knowledge Sharing 11
Creating Predictive Models using ML Studio ML Studio along with the assortment of modules it offers for modeling workflow, we can often build sophisticated models without writing a single line of code. It also enables us to insert R and Python code anywhere in the workflow, providing infinite flexibility in what you can model. Machine Learning starts with data, which can come from a variety of sources. Once the data is ready, you select an algorithm and train the model by allowing it to iterate over the data and find patterns in it. After that comes scoring and evaluating the model, which tells you how well the model is able to predict outcomes. All of this is performed visually in ML Studio. Once the model is ready, a few button clicks deploy it as a Web service so it can be called from client apps. 2018 Dell EMC Proven Professional Knowledge Sharing 12
ML Studio provides canned implementations of 25 of the classic algorithms used in machine learning. It divides them into four categories. Anomaly detection is the identification of items, events, or observations which do not conform to an expected pattern or other items in a dataset. A classic example is examining a dataset representing banking transactions and detecting potentially fraudulent transactions in that group. Regression algorithms seek to establish and quantify relationships between variables. By establishing a relationship between a dependent variable and one or more independent variables, regression analysis can enable the value of a dependent variable to be predicted given a set of inputs with quantifiable accuracy. 2018 Dell EMC Proven Professional Knowledge Sharing 13
The purpose of classification algorithms is to identify the category to which an observation belongs based on training data consisting of observations which have already been classified (assigned to a category). Clustering seeks to group a set of objects in such a way that objects in the same group (called a cluster) are more similar to each other than to those in other groups (clusters). The Azure ML Cheat Sheet helps you pick the right algorithm for a model, even if you re not a trained data scientist. One example is if you want to use a set of input values to predict an output value from a continuous set of values (e.g. a person s age), use linear regression. But if you re more interested in the distribution of the output, you might use fast forest quantile regression instead. Classification algorithms, by contrast, are used to predict a value from a discrete set of values for example, classifying an e-mail as spam or not spam. Once deployed as a Web service, a model can be used with simple REST calls over HTTP. This enables developers to build smart apps that get their intelligence from ML. 2018 Dell EMC Proven Professional Knowledge Sharing 14
Language Understanding Intelligence Service Language Understanding (LUIS) allows your application to understand what a person wants in their own words. LUIS uses machine learning to allow developers to build applications that can receive user input in natural language and extract meaning from it. A client application that converses with the user can pass user input to a LUIS app and receive relevant, detailed information back. What is a LUIS app? A LUIS app is a domain-specific language model designed by you and tailored to your needs. You can start with a prebuilt domain model, build your own, or blend pieces of a prebuilt domain with your own custom information. A model starts with a list of general user intentions such as "Book Flight" or "Contact Help Desk." Once the intentions are identified, you supply example phrases called utterances for the intents. Then you label the utterances with any specific details you want LUIS to pull out of the utterance. Prebuilt domain models include all these pieces for you and are a great way to start using LUIS quickly. After the model is designed, trained, and published, it is ready to receive and process utterances. The LUIS app receives the utterance as an HTTP request and responds with extracted user intentions. Your client application sends the utterance and receives LUIS's evaluation as a JSON object. Your client app can then take appropriate action. Key LUIS concepts Intents: Intent represents actions the user wants to perform. The intent is a purpose or goal expressed in a user's input, such as booking a flight, paying a bill, or finding a news 2018 Dell EMC Proven Professional Knowledge Sharing 15
article. You define and name intents that correspond to these actions. A travel app may define an intent named "Book Flight." Utterances: An utterance is text input from the user that your app needs to understand. It may be a sentence, like "Book a ticket to Paris", or a fragment of a sentence, like "Booking" or "Paris flight." Utterances aren't always well-formed, and there can be many utterance variations for a particular intent. Entities: An entity represents detailed information that is relevant in the utterance. For example, in the utterance "Book a ticket to Paris", "Paris" is a location. By recognizing and labeling the entities that are mentioned in the user s utterance, LUIS helps you choose the specific action to take to answer a user's request. Accessing LUIS LUIS has two ways to build a model: the Authoring APIs and the LUIS.ai web app. Both methods give you and your collaborators control of your LUIS model definition. We can use either LUIS.ai or the Authoring APIs or a combination of both to build your model. This management includes models, versions, collaborators, external APIs, testing, and training. Once the model is built and published, we pass the utterance to LUIS and receive the JSON object results with the Endpoint APIs. Author LUIS model Begin the LUIS model with the intents the client app can resolve. Intents are just names such as "BookFlight" or "OrderPizza." After an intent is identified, we need sample utterances that we need LUIS to map to our intent such as "Buy a ticket to Seattle tomorrow." Then, label the parts of the utterance that are relevant to our app domain as entities and set a type such as date or location. Generally, an intent is used to trigger an action and an entity is used as a parameter to execute an action. For example, BookFlight" intent could trigger an API call to an external service for booking a plane ticket, which requires entities like the travel destination, date, and airline. 2018 Dell EMC Proven Professional Knowledge Sharing 16
Identify Entities Entity identification determines how successfully the end user gets the correct answer. LUIS provides several ways to identify and categorize entities. Prebuilt Entities: LUIS has many prebuilt domain models including intents, utterances, and prebuilt entities. You can use the prebuilt entities without having to use the intents and utterances of the prebuilt model. The prebuilt entities save you time. Custom Entities: LUIS gives you several ways to identify your own custom entities including simple entities, composite entities, list entities, and hierarchical entities. Phrases: LUIS provides phrase lists, which also help identify entities. Improve Performance Once our application is published and real user utterances are entered, LUIS uses active learning to improve identification. In the active learning process, LUIS provides real utterances that it is relatively unsure of for you to review. You can label them according to intent and entities, retrain, and republish. This iterative process has tremendous advantages. LUIS knows what it is unsure of, and our help leads to the maximum improvement in system performance. LUIS learns quicker, and takes the minimum amount of our time and effort. LUIS is active machine learning at its best. The Microsoft Bot Framework Microsoft bot framework provides a platform for building and publishing bots. You can use the bot builder SDK or the Azure bot service to create your bot and publish it as a web service then make your bot available through one or more channels such as Skype or other popular messaging and social media platforms. The bot connector service handles the message exchange between your bot and the channels through which users engage with it. Each interaction is an activity that can be a message or a visual element such as a card that contains images buttons or input field. When a user starts a dialogue with a bot, the activities are routed between the channel and the bot automatically. You can also integrate your bot with cognitive services so that it can respond intelligently to user input. 2018 Dell EMC Proven Professional Knowledge Sharing 17
Cognitive Toolkit Cognitive Toolkit (CNTK) will enable enterprise-ready, production-grade AI by allowing users to create, train, and evaluate their own neural networks that can then scale efficiently across multiple GPUs and multiple machines on massive data sets. CNTK is a framework for describing learning machines. Although intended for neural networks, the learning machines are arbitrary in that the logic of the machine is described by a series of computational steps in a Computational Network. CNTK can be included as a library in your Python, C#, or C++ programs. Additionally, you can use the CNTK model evaluation functionality from your Java program. With support for Keras, users will now benefit from the performance of CNTK without any changes to their existing Keras recipes. Computational Network defines the function to be learned as a directed graph where each leaf node consists of an input value or parameter, and each non-leaf node represents a matrix or tensor operation upon its children. The beauty of Cognitive Toolkit is that once a computational network has been described, all the computation required to learn the network parameters are taken care of automatically. There is no need to derive gradients analytically or to code the interactions between variables for backpropagation. 2018 Dell EMC Proven Professional Knowledge Sharing 18
Conclusion The work of understanding our responsibilities in developing and deploying safe and ethical AI systems is ongoing. Development of trust will come through use over time, just as trust was built with all technologies that preceded AI, and all that will follow it. Compose intelligent applications, customized to your organization s availability, security, and compliance requirements with Microsoft AI platform. With the Azure platform and productivity services, you can create the next generation of applications that span an intelligent cloud and an intelligent edge powered by AI. Use a comprehensive set of flexible AI Services for any scenario, enterprise-grade AI Infrastructure that run AI workloads anywhere at scale, and modern AI Tools for developers and data scientists to create AI solutions easily and with maximum productivity. References 1. Microsoft Azure Notebooks: https://notebooks.azure.com/ 2. Microsoft Cognitive Toolkit: https://www.microsoft.com/en-us/cognitive-toolkit/ 3. Azure Machine Learning Studio: https://azure.microsoft.com/en-us/services/machinelearning/ 4. Keras: https://keras.io/ Dell EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without notice. THE INFORMATION IN THIS PUBLICATION IS PROVIDED AS IS. DELL EMC MAKES NO RESPRESENTATIONS OR WARRANTIES OF ANY KIND WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION, AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Use, copying and distribution of any Dell EMC software described in this publication requires an applicable software license. Dell, EMC and other trademarks are trademarks of Dell Inc. or its subsidiaries. 2018 Dell EMC Proven Professional Knowledge Sharing 19