By Nicola Wright
Is your business taking advantaging of artificial intelligence and machine learning? Because your competitors are.
Whether you build your own AI-fuelled apps or use a ready-made service like Microsoft Azure Machine Learning, artificial intelligence is the new normal for small businesses and enterprises alike.
Advances in data science and automation have made technology that may have seemed within reach of only the biggest, techiest companies accessible to businesses of all shapes and sizes.
Today, more and more organizations are leveraging AI, machine learning, and deep learning to power better service, smarter forecasting, and more efficient operations—a massive 61% of businesses said they implemented AI in 2017, a significant rise from the 38% who did so in 2016.
With the likes of predictive analytics, chatbots, and natural language processing now available as a service, it’s easier than ever for companies to get on board with this revolutionary tech. If you’re not sure where to start, we’ve answered some of the most common questions about machine learning on Microsoft Azure.
Read on, and get ready to welcome our new corporate robot overlords.
Machine learning is a technique that allows computers to read and absorb data, and use that information to make predictions without being explicitly “taught” how to do so.
Once they’ve digested existing information, devices and apps that feature machine learning technology can make forecasts about customer behavior, situational outcomes, and future trends.
AI offers a huge number of possibilities for businesses, with organizations already using machine learning to streamline operations, improve customer experience, and boost profits across every industry.
Even if you’re not quite there with your own business, you almost definitely will have had first-hand experience of corporations using machine learning. Retailers can use machine learning to up- and cross-sell by looking at what a customer has already bought, and predicting what products they might also be interested in.
Netflix does something similar, examining the content you watch and suggesting things you might also like, based on the oceans of user behaviour data is has at its disposal. The aim of these algorithms is to increase the amount of time you spend using the app, and give you a better experience, and they’re hugely successful; more than 80% of what Netflix users watch is content that’s been suggested by the app.
Uber uses data about traffic, weather conditions, and nearby events to automatically set its ever-fluctuating price levels, Lidl recently introduced a conversational chatbot to help customers select and pair its wines, and AI has even reached the agricultural sector. John Deere uses machine learning to teach its robots to autonomously discern which plants are pests, and the best pesticide to use to treat them.
One of the great benefits of infusing an app or website with AI and machine learning is that it will constantly enhance itself. The more data a machine learning platform is fed, the more its “brain” grows, develops, and becomes better at what it does. Machine learning is voracious by nature, and never stops improving.
Like most apps and digital services today, machine learning is available as a service—meaning that users can access them via the internet without needing any software installed on their own machines.
Using as a service platforms means that customers get the processing power of the vendors that are building and hosting them without having to host or maintain and the software themselves.
This backing by big-name tech giants is a crucial part of what’s making machine learning as a service so accessible. Machine learning as a service products are designed to help users get up to speed with machine learning fast, even if they don’t have much data science expertise.
There’s a huge range of products and services that sit under the MLaaS label, both fully- and semi-automated, including those that handle data pre-processing, model training, and text translation.
As you’d expect from the world’s biggest cloud service provider, Microsoft offers an extensive selection of AI and machine learning services on its Azure platform. Let’s take a look at what’s available.
Azure Machine Learning Service is a cloud-based service that allows users to create, teach, launch, and manage their own machine learning models on any scale.
Customers can take advantage of Azure’s powerful machine learning platform as a foundation to make building their own ML solutions quicker and easier.
The service gives users access to automated machine learning, making identifying the best algorithms and configuring hyperparameters much faster, and improving productivity and reducing costs through autoscaling.
Azure Machine Learning Service users can build solutions using a range of open-source frameworks, like PyTorch, TensorFlow and scikit-learn, and then can deploy them both to the cloud and the edge.
You can start small, beginning by training your solution on your local machine and then scaling it out on Azure’s cloud platform.
It’s compatible with container services—like Docker, Azure Container Instances, or Azure Kubernetes Service—too, so it’s easy to transport your ML solution to wherever it’s needed.
Azure Machine Learning Service is an end-to-end ML solution, enabling users to manage and track their models after deployment, make multiple runs to find the best solution, and return predictions in real-time.
The product’s automation features mean that you don’t have to be a developer and data scientist to use it, making it accessible to smaller businesses who want to leverage machine learning despite not having a dedicated in-house data science team.
Veronika Kolesnikova is a Microsoft MVP, and has provided her insights into how you can utilize Azure Functions for machine learning.
“Azure Functions open the whole new world of serverless. They are used as a tool to “run small pieces of code in the cloud” and can be used for all kinds of tasks: data export and import from APIs, databases and files, accessing applications hosted in a cloud by a trigger or accessing, updating, creating ML models, etc.
“Azure Functions support some of the most popular languages for ML: Python, Java and C#. So it allows you to use, for example, Python and TensorFlow with a machine learning model for different scenarios: image recognition, sentiment analysis, price prediction, etc. It’s easy to create a function from Azure portal or right from Visual Studio, so there’s no additional set up required.
“There’re so many options available for using Azure Functions for ML tasks:
– Import/export data for your model from a database or API endpoint
– Calling one (or more) Cognitive Services API for analyzing data
– Accessing custom pre-built models in a cloud or server to analyze data, etc.
“Let me provide an example of the third option above. I’m currently using ML.NET for some of my personal projects. It became available for all .NET developers in May 2019, but it has been used inside Microsoft in Office 365, Power BI and other tools for quite some time. It’s very appealing to me because I can continue using C# (or F#) for building and using custom ML models.
“Recently I decided to use ML.NET in my Xamarin application to work with a custom model I created. Unfortunately currently ML.NET is not supported on ARM processor architecture, so Xamarin apps (iOS, Android) and ARM-based IoT devices are out of luck. There’re several available workarounds, including creating an Azure Function that will call ML.NET application, pass data and get the result from it. The function can be triggered from the Xamarin application, so it’ll get the result without accessing the ML.NET application directly (another workaround is mentioned in a blog post I wrote.)
“As you can see the world of ML is endless and exciting. You don’t have to be a scientist anymore to use all the benefits of it in your applications. There’re lots of tools including Azure Functions that will make learning and implementation processes much more accessible. They will open the door to that exciting world for you.”
Not be confused with Machine Learning Service, Azure Machine Learning Studio is a visual workspace which lets users create machine learning solutions using a drag–and–drop system, with no coding required.
Whereas Machine Learning Service is a managed cloud service for building, training, and deploying machine learning models using Python and CLI, Machine Learning Studio is a simpler platform. Machine Learning Studio is browser-based, allowing users to click their way from initial idea to deployment using Azure’s prebuilt and preconfigured algorithms and data modules.
Users can collaborate in Machine Learning Studio every step of the way, from building and testing to deploying their solutions. Though built with ease of use in mind, Machine Learning Studio is no less powerful than its code-based sibling, boasting the most comprehensive set of machine learning tools offered by a major cloud provider, and far outdoing those available through Amazon, Google, and IBM.
Because it comes pre-loaded with algorithms, users can build and deploy machine learning solutions super quickly; provided, that is, that the preconfigured algorithms are adequate for your model. If you want to create your own from scratch, you’ll need Machine Learning Service.
“Hi! I’m Daniel. How can I help you today?”
If you’ve visited a website recently and had a little box pop up in the corner as some person, animal, or anthropomorphic mascot offers you its services, you’ve already had an encounter with a chatbot.
Chatbots have exploded in popularity in recent years, thanks in no small part to the fact that they offer a fast, convenient way for customers to connect with businesses and get information at any hour, without languishing in a phone queue. Powered by AI, chatbots can help transform customer experience by resolving small issues quickly and easily
Boasting the ability to surface information, answer simple queries, and replicate natural human conversation, chatbots are already being put to good use by businesses of all types and sizes. And with the artificial intelligence that powers these bots becoming smarter and more accessible by the day, they aren’t going away any time soon.
The machine learning algorithms under the hood of a chatbot program enable it to understand patterns, respond appropriately, and continuously learn from its exchanges to improve future interactions. And they aren’t just useful for websites. Chatbots can also be injected into third-party sources like Skype, Slack, and Facebook so customers can access them on whichever platform they’re using.
Microsoft has built its own platform on which customers can build and deploy chatbots, taking advantage of the power, reliability, and support that comes with utilising chatbots as a service.
Azure Bot Service allows customers to create bots quickly, manage them, and launch them across a variety of platforms on a pay-as-you-go basis.
Azure’s bot framework gives users the tools to build a huge range of chatbots for a variety of uses—the service offers many out-of-the-box templates for scenarios including language understanding, question and answer, and proactive bots.
They can be used to autonomously answer customer questions, surface information from a connected CRM or calendar apps, recognise users in photos, translate language, moderate content, and make smart, personalized recommendations.
Azure bots can also be integrated with Azure Cognitive Services to add the ability to understand natural language and images, and customized for various industry-specific situations like banking, travel, and entertainment.
European phone retailer Dixons Carphone used Microsoft Bot Framework and Microsoft Cognitive Services to create their customer chatbot, Cami.
Deployed on both the brand’s website and Facebook Messenger, Cami answers customer queries and allows both customers and in-store employees to research, locate, and save products, as well as check stock.
On top of being able to read and understand text-based input, Cami is also able to use pictures of in-store shelf labels to check stock status across multiple stores.
Azure Cognitive Services are a set of pre-trained, ready to consume “smart” products. Cognitive Services allows customers to fortify their platforms with intelligent algorithms without having to build them from scratch. These algorithms then enable apps, bots, and websites to see, hear, speak, understand and interpret user needs in a way that feels natural and human.
The capabilities offered through Cognitive Services can be broken down into five areas:
These services help customers monitor performance and content, and personalize experiences for your users. Essentially, Decision services shoulder some administrative burden by making informed and efficient decisions on your behalf.
That might mean moderating content to spot anything that doesn’t meet your standards (offensive language, images, or video content for example) and filtering it out with the Content Moderator service. You could use these services to examine the health of your business in real time, or oversee IoT devices remotely using Anomaly Detector. You can also deliver customized user experiences within your apps with Personaliser and its reinforcement-learning cycle that constantly absorbs new information about your users.
People communicate using many methods, and Azure’s Vision services make it possible for your apps to recognize and analyze content not only in text form, but also in forms, images, video, and digital ink.
The Computer Vision service is able to identify people and landmarks, classify images, and perform optical character recognition to read non-digitized information.
The Face service detects and identifies not only people, but emotions in images, and is able to group them accordingly.
Video Indexer can do the same thing in video, and can detect objects, scenes, and specified activities within film.
The Form and Ink Recogniser services can read and extract content from forms and digital handwriting.
Azure’s Speech services allow customers to integrate speech processing capacities into any app or service. Currently, there are two speech-based facilities available. Speech Services can automatically transcribe speech into text, turn text into speech using customizable voice options, and translate languages in real time. Speaker Recognition, on the other hand, identifies and verifies speakers.
Search services help bring smart, powerful web-scale search muscle to apps and websites, ad-free. Azure offers a range of specialized search services, from web, news, and local business search, to media-focused tools like video, image, and visual search.
Azure’s Language tools are built to allow the processing of natural, human language—these services enable users to evaluate customer sentiment and recognize exactly what customers want, so you can decide how best to help them.
Understand language in context, analyze feelings and extract key phrases, detect and translate languages, and create knowledge bases to allow bots to unearth questions and offer answers based on unstructured text.
Created in partnership with the makers of Apache Spark, Azure Databricks is an analytics platform, built to give customers access to powerful data analytics through a user-friendly platform.
Optimized for Microsoft Azure cloud services, Databricks is an interactive workspace that’s easy to set up, and can be used to streamline analytical workflows and help data professionals collaborate.
Databricks is infused with AI, powered by Azure Cognitive Services, which enables users to supercharge their data analysis and uncover valuable insights using smart algorithms to spot patterns and trends. Data is captured, stored, and processed in real-time, and insights are surfaced through analytical dashboards.
Users can build, train, and deploy their own AI models using Databricks to meet custom deep learning needs, or they can utilize one of the platform’s many pre-configured frameworks and can get started right away.
Databricks is pretty flexible; it can be used with a range of languages and deep learning frameworks, and can be scaled up or down automatically to meet changing processing and machine learning requirements.
As with any service on Azure, it’s near impossible to give a straightforward answer on pricing.
Azure’s various AI and machine learning services are priced based on various factors, like where you’re located, how much computing power you use, how much storage space you need, and what other Azure products you have to utilize to in order to create, deploy, and run the apps and services you need.
Azure’s pricing calculator should give you a rough idea of the kind of costs you’ll be looking at for its AI and machine learning services.
Browse pre-qualified candidates now and find the talent you need to transform your business.
Resgistrate para recibir consejos sobre Microsoft Azure y Dynamics