Machine Learning as a Service (MLaaS)

There has been a paradigm shift in the way that businesses build their technology stacks in recent years driven by a major move into digital platforms and microservices. This shift has been made possible because of the cloud computing revolution, particularly the massive growth of public cloud services provided by enterprise companies such as Amazon, Microsoft and Google, among others. These enterprises have put an enormous emphasis on the “as a service” business model, which allows outside companies to pick and choose necessary microservices provided by the enterprises.

Data is the driver behind machine learning.

Infrastructure as a service (IaaS) and cloud platform as a service (PaaS) are the two most commonly used service offerings; however, in the coming year businesses will rapidly begin to adopt machine learning as a service (MLaaS) into their technology stacks for a number of reasons, the main one being the need to progress company-wide digital transformation.

By adopting artificial intelligence software and services, businesses can enhance product capabilities, better interact with customers, streamline business operations, and create predictive and precise business strategies. Developers can build quickly and efficiently with MLaaS offerings, because they have access to pre-built algorithms and models that would take them extensive resources to build otherwise. The developers who have the knowledge and skills to build machine learning models are few and far between, and are very expensive, so the ease and speed of setup coupled with the monetary benefits will be a major draw for business implementing AI in 2018.

Data is the driver behind machine learning, and because these huge companies produce and have access to so much data, they are able to build and train their own machine learning models in house. This allows them to offer it to outside companies as MLaaS, the same way that since they have more datacenter space than smaller companies they can provide IaaS. Generally, smaller companies do not have access to as much data to create powerful AI models; however, they do have valuable data that can be fed to pre-trained machine learning algorithms to create business-critical outcomes or actionable insights.

There are a number of MLaaS offerings for businesses to choose from, including natural language processing (NLP), computer vision, AI platforms and other machine learning APIs. Amazon, Google, Microsoft and IBM all offer different services for these machine learning functionalities. Additionally, these different types of AI can have unique impacts on many aspects of digital transformation.


How Amazon revolutionized modern business technologies

Amazon was the first major enterprise company to offer IaaS and PaaS, and it has taken that business model and applied it to all the technologies needed to scale a business. Amazon Web Services (AWS) now has a large catalog of microservices that companies can purchase to create their own digital platforms. The concept of digital platforms and microservices are extensive, but AWS, along with Microsoft’s Azure and Google’s Cloud Platform, offer everything from cloud computing, storage and database management, to augmented and virtual reality, business productivity applications, and tools for the internet of things. These microservices are largely API-based so they make for a quick deployment, which is largely the appeal of these products. It is challenging and time consuming to build a development environment if you are trying to rapidly scale or build a product.

The last point is something that Amazon as a company came to realize early on when it was trying to build new in-house applications. According to Andy Jassy, AAWS’ CEO, around the year 2000, before Amazon was the dominant force it is today, it hoped to build an external development platform called Merchant.com to assist other merchants (such as Target) in building their own online shopping sites atop Amazon’s e-commerce engine. However, in order to build the external development environment, Amazon needed to clean up its internal environment.

As many rapidly growing companies do, Amazon did not build its in-house solutions to scale, so it began to reorganize its development environment into a series of APIs that could then be used by development teams across the company. As teams built out easy-to-understand documentation around the APIs, it became simple to share internally, and increase the speed and ease of developing new applications for Amazon, which helped Amazon grow faster and more efficiently. This concept is the same as it is today for AWS’ microservices that can be purchased by third-party companies.

Businesses will greatly lean into MLaaS in 2018, and enterprise services will be the reason why.

Similarly, Amazon found that it was struggling with siloed infrastructure throughout the development team, which made projects lengthier than they needed to be. “When Jassy, who was Amazon CEO Jeff Bezos’ chief of staff at the time, dug into the problem, he found a running complaint. The executive team expected a project to take three months, but it was taking three months just to build the database, compute or storage component. Everyone was building their own resources for an individual project, with no thought to scale or reuse.”

To solve this issue, Amazon sought to build a singular internal infrastructure service that all development teams could have access to so there would be no need to build their own every single time.

This infrastructure necessity forced Amazon into building fast-scaling, reliable data centers that didn’t cost too much money to maintain. As it grew and compiled more and more data centers, Amazon began to realize that outside companies could run its applications on top of Amazon’s infrastructure. Some three years later, in August 2006, it released Amazon Elastic Compute Cloud, the first IaaS to hit the market. It took some time for public cloud infrastructure to make its mark in the modern business world, but now the likes of Microsoft, Google and IBM are all still playing catch up, some 12 years later, to gain their own chunk of market share.

It just so happens that these cloud providers are working equally as hard at being at the forefront of AI advancements, and have built machine learning models that they are providing as microservices to outside businesses. Similarly to the way that the other cloud services have grown, and continue to grow rapidly, businesses will greatly lean into MLaaS in 2018, and these enterprise services will be the reason why. There is an opportunity out there to grab market share, similarly to the way that Amazon was able to take hold of the IaaS market, and Amazon, Microsoft, Google,and IBM will fight, scrape and claw for it. In coming years, it will be necessary to include MLaaS as a chunk of the cloud services revenue.


Why MLaaS from enterprise companies will be the driver of AI digital transformation

Any time there are a few big companies in a space it appears ripe for disruption, but enterprises are poised to dominate the MLaaS space for a variety of reasons, the first being big data. Simply put, they have it. It is getting easier to access open data sets (which are often open-sourced by the enterprise companies), but these corporations have access to exponentially more data than small or mid-sized businesses. Because they have data they have been able to build machine learning algorithms and train them with said data. This is an advantage that is almost insurmountable for small competitors and startups.

Another major advantage is competitive salaries the likes of Amazon, Microsoft, Google and IBM can offer AI developers, of which there are not many. Lean startups cannot afford to pay even close to a similar salary, and offering a stake in ownership is no longer enough. According to a New York Times piece from October 2017, large tech companies have put such an emphasis on AI that they are willing to pay beyond top dollar. “Typical A.I. specialists, including both Ph.D.s fresh out of school and people with less education and just a few years of experience, can be paid from $300,000 to $500,000 a year or more in salary and company stock, according to nine people who work for major tech companies or have entertained job offers from them,” the article read.

“In the entire world, fewer than 10,000 people have the skills necessary to tackle serious artificial intelligence research.”Element AI

Now some might pass up those salaries to build something on their own or join a startup where they can have more control, but those unguaranteed common stock shares are difficult to compare to those shares of the tech giants. “Well-known names in the A.I. field have received compensation in salary and shares in a company’s stock that total single- or double-digit millions over a four- or five-year period,” read the Times’ piece. “And at some point they renew or negotiate a new contract, much like a professional athlete.”

Remind me again why I studied creative writing?

The ability to grab talent is so important because there isn’t much of it. There is a major lack of people who are knowledgeable and skilled enough to build AI applications. Per the New York Times article, “In the entire world, fewer than 10,000 people have the skills necessary to tackle serious artificial intelligence research,” according to Element AI, an independent lab in Montreal, per the New York Times article.

These jobs are relatively new, so while there are many data science and machine learning courses being offered to students, it still takes years to receive the education needed to develop AI, so the talent gap will continue for some time. Since enterprise companies have the resources to attract the talent, they are able to build out their machine learning services, which will only increase the need for other businesses who can’t afford those employees to utilize MLaaS.

Also, many businesses already take advantage of public cloud providers, so adding one more microservice from the catalog is not too much of a hassle. If a business is already storing its data in an AWS or Azure public cloud, it is easy to adopt an MLaaS solution from those vendors. They can work with a business’ data, which is stored on their infrastructure, and help train their machine learning service to benefit the business. Not only will it be a quick deployment, but most likely inexpensive, another draw of microservices in general. This MLaaS approach will save the business time, energy and resources (which they do not have enough of), to help modernize a business with AI.


MLaaS products available to businesses

Enterprises have already released a number of machine learning services for use by outside businesses. These MLaaS offerings consist of AI platforms for custom algorithm building, natural language processing, speech recognition, computer vision and other miscellaneous machine learning models. Below are MLaaS offerings from AWS, Microsoft Azure and Google Cloud Platform.

AI platforms

Artificial intelligence software platforms allow users to build and train machine and deep learning models and applications. These solutions are similar to cloud platforms that help you build applications, in the sense that they often utilize drag-and-drop functionality for easy construction of algorithms and models. Users can pump data through these solutions to best train their models to perform the tasks they need. The following MLaaS offerings are AI platforms.

  • Release Date: November 2017
  • Description: Amazon SageMaker is a fully-managed service that enables developers and data scientists to quickly and easily build, train, and deploy machine learning models at any scale. Amazon SageMaker removes all the barriers that typically slow down developers who want to use machine learning.
  • Release Date: February 2015
  • Description:Studio is a powerfully simple browser-based, visual drag-and-drop authoring environment where no coding is necessary. Go from idea to deployment in a matter of clicks.”
  • Review Excerpts:
    • Likes: “It has got a nice user interface, drag and drop components and connect them to build your model. You can also customize various components or code in Python or R to have a custom component. It has nice interface to visualize data and easy to deploy your project. You can upload data sets from various sources. The best thing is its amazing interface. It also has Cortana intelligence suite and other stuff worth trying. It gets work done fast and easy. It trains the model fast.”
    • Likes: “I like that people that don’t have a lot of programming experience in languages like R or Python can use ‘templates’ for running machine learning algorithms. This makes this tool available for non-experts in the field.”
  • Release Date: March 2016 (Replaced Google Prediction API)
  • Description: Google Cloud Machine Learning Engine is a managed service that enables you to easily build machine learning models that work on any type of data, of any size. Create your model with the powerful TensorFlow framework that powers many Google products, from Google Photos to Google Cloud Speech.
  • Review Excerpt:
    • Likes: “Very easy to sign up and use REST API and different language SDKs.”
    • Business Problems Solved: “Image analysis (i.e., multi-object, OCR). The accuracy and performance of Google Cloud ML services are amazing.”

Natural language processing

Natural language processing (NLP) is a form of AI that allows applications to interact with human language. NLP has made major strides in recent years due to the rapid advancement of deep learning, particularly with the utilization of artificial deep neural networks. The applications of NLP include machine translation, grammar parsing, sentiment analysis and part-of-speech tagging, among other uses. NLP is the primary technology behind tools like chatbots. The following MLaaS offerings are NLP solutions.

  • Release Date: November 2016
  • Description:Amazon Polly is a service that turns text into lifelike speech, allowing you to create applications that talk, and build entirely new categories of speech-enabled products. Amazon Polly is a text-to-speech service that uses advanced deep learning technologies to synthesize speech that sounds like a human voice.”
  • Release Date: April 2017
  • Description:Amazon Lex is a service for building conversational interfaces into any application using voice and text. Amazon Lex provides the advanced deep learning functionalities of automatic speech recognition (ASR) for converting speech to text, and natural language understanding (NLU) to recognize the intent of the text, to enable you to build applications with highly engaging user experiences and lifelike conversational interactions.”
  • Release Date: November 2017
  • Description: Amazon Translate is a neural machine translation service that delivers fast, high-quality, and affordable language translation. Neural machine translation is a form of language translation automation that uses machine learning and deep learning models to deliver more accurate and more natural sounding translation than traditional statistical and rule-based translation algorithms.”
  • Release Date: April 2017
  • Description: Amazon Comprehend is a natural language processing (NLP) service that uses machine learning to find insights and relationships in text. Amazon Comprehend identifies the language of the text; extracts key phrases, places, people, brands, or events; understands how positive or negative the text is; and automatically organizes a collection of text files by topic.”
  • Release Date: December 2017 (out of beta)
  • Description: Azure Language Understanding Intelligent Service (LUIS) is a service that enable user to quickly deploy an HTTP endpoint that will take the sentences being send and interpret them in terms of the intention they convey and the key entities that are present, it has a web interface that can custom design a set of intentions and entities that are relevant to an application and guide ser through the process of building a language understanding system.”
  • Release Date: July 2016 (out of beta)
  • Description: Azure Bing Spell Check API is a tool that help users correct spelling errors, recognize the difference among names, brand names, and slang, as well as understand homophones as they’re typing.”
  • Release Date: August 2016
  • Description: Microsoft Web Language Model API is a REST-based cloud service that provide tools for natural language processing, using this API, users application can leverage the power of big data through language models trained on web-scale corpora collected by Bing in the EN-US market.”
  • Release Date: September 2017 (out of beta)
  • Description: Microsoft Text Analytics API is a suite of text analytics services that offer APIs for sentiment analysis, key phrase extraction and topic detection for English text, as well as language detection for 120 languages.”
  • Release Date: October 2016
  • Description: Microsoft Linguistic Analysis APIs is a tool that provide access to natural language processing (NLP) that identify the structure of text and it provides three types of analysis: Sentence separation and tokenization, Part-of-speech tagging and Constituency parsing.”
  • Release Date: November 2016
  • Description: Microsoft Translator Text API is a cloud-based machine translation service supporting multiple languages that reach more than 95% of world’s gross domestic product. Translator can be used to build applications, websites, tools, or any solution requiring multilanguage support.”
  • Release Date: Release Date: April 2006 (as statistical machine translation) and November 2016 (as neural machine translation)
  • Description: Cloud Translation API provides a simple programmatic interface for translating an arbitrary string into any supported language using state-of-the-art Neural Machine Translation. Translation API is highly responsive, so websites and applications can integrate with Translation API for fast, dynamic translation of source text from the source language to a target language (e.g., French to English).”
  • Review Excerpt:
    • Likes: “Updated to use TensorFlow Recurrent neural networks for translation, so you get Machine Translation with very competitive BLEU scores for select (mostly European) languages.”
    • Business Problems Solved: “Website translation for dynamic content / intermediate translation for pending translation items (programmatically via API)”
  • Release Date: November 2016
  • Description:Google Cloud Natural Language API reveals the structure and meaning of text by offering machine learning models in an easy to use REST API, user can use it to extract information about people, places, events and much more, mentioned in text documents, news articles or blog posts and to understand sentiment about product on social media or parse intent from customer conversations happening in a call center or a messaging app.”

Speech recognition

Speech recognition software models allow applications to understand spoken language and convert conversations into text. Most often, developers combine speech recognition with natural language understanding and intent analysis to determine what a user wants. The most familiar example of this technology in action is with Apple’s Siri, Google’s Home and Amazon’s Alexa; however, the following MLaaS speech recognition tools can be used for a variety of purposes:

See above
  • Release Date: November 2017
  • Description:Amazon Transcribe is an automatic speech recognition (ASR) service that makes it easy for developers to add speech to text capability to their applications. Using the Amazon Transcribe API, you can analyze audio files stored in Amazon S3 and have the service return a text file of the transcribed speech.”
  • Release Date: September 2017
  • Description:Microsoft Bing Speech API is a cloud-based API that provides advanced algorithms to process spoken language, it allow developers add speech driven actions to their applications including real-time interaction with the user.”
  • Release Date: March 2016
  • Description:Microsoft Speaker Recognition API is a cloud-based APIs that provide the most advanced algorithms for speaker verification and speaker identification that can be divided into two categories: speaker verification and speaker identification.”
  • Release Date: March 2016
  • Description: Microsoft Translator Speech API is a cloud-based automatic translation service. The API enables developers to add end-to-end, real-time, speech translations to their applications or services.”
  • Release Date: February 2017
  • Description:Custom Speech Service is a cloud-based service that provides users with the ability to customize speech models for Speech-to-Text transcription.”
  • Release Date: April 2017 (out of beta)
  • Description:Google Cloud Speech API is a service that enables developers to convert audio to text by applying neural network models in an easy to use API, it recognizes over 80 languages and variants, to support global user base and can transcribe the text of users dictating to an application’s microphone, enable command-and-control through voice, or transcribe audio files, among many other use cases.”
  • Release Date: November 2017
  • Description:Dialogflow is an end-to-end development suite for building conversational interfaces for websites, mobile applications, popular messaging platforms, and IoT devices. You can use it to build interfaces (e.g., chatbots) that are capable of natural and rich interactions between your users and your business. It is powered by machine learning to recognize the intent and context of what a user says, allowing your conversational interface to provide highly efficient and accurate responses.”

Computer vision

Computer vision is the deep learning technology that allows applications to recognize and understand images and videos. These solutions can analyze visual content to identify and classify specific objects, people and streaming video. Computer vision algorithms can help companies such as YouTube ensure that there is not inappropriate content being uploaded onto their sites without requiring a human view every video. The following MLaaS offering provide computer vision:

  • Release Date: November 2016
  • Description:Amazon Rekognition makes it easy to add image and video analysis to your applications. It can identify the objects, people, text, scenes, and activities, or any inappropriate content from an image or video.”
  • Release Date: April 2017 (out of beta)
  • Description: “The cloud-based Computer Vision API provides developers with access to advanced algorithms for processing images and returning information. By uploading an image or specifying an image URL, Microsoft Computer Vision algorithms can analyze visual content in different ways based on inputs and user choices.”
  • Release: April 2017 (out of beta)
  • Description:Content moderation is the process of monitoring user-generated content on online and social media websites, chat and messaging platforms, enterprise environments, gaming platforms, and peer communication platforms. The goal is to track, flag, assess, and filter out offensive and unwanted content that creates risk for your organization. Moderated content might include text, images, and videos.”
  • Release Date: May 2017
  • Description: Custom Vision Service is a tool for building custom image classifiers. It makes it easy and fast to build, deploy, and improve an image classifier. We provide a REST API and a web interface to upload your images and train.”
  • Release Date: April 2017 (out of beta)
  • Description: “Microsoft Face API is a cloud-based service that provides the most advanced face algorithms. Face API has two main functions: face detection with attributes and face recognition.”
  • Release Date: February 2017
  • Description: “The Emotion API takes a facial expression in an image as an input, and returns the confidence across a set of emotions for each face in the image, as well as bounding box for the face, using the Face API. If a user has already called the Face API, they can submit the face rectangle as an optional input.”
  • Release Date: May 2017
  • Description: Video Indexer is a cloud service that enables you to extract the following insights from your videos.”
  • Release Date: February 2016
  • Description:Google Cloud Vision API is a tool that enables developers to understand the content of an image by encapsulating powerful machine learning models in an easy to use REST API, it quickly classifies images into thousands of categories (e.g., “sailboat”, “lion”, “Eiffel Tower”), detects individual objects and faces within images, and finds and reads printed words contained within images user can build metadata on image catalog, moderate offensive content, or enable new marketing scenarios through image sentiment analysis.”
  • Release Date: March 2017
  • Description: Google Cloud Video Intelligence API makes videos searchable, and discoverable, by extracting metadata with an easy to use REST API.”

Other machine and deep learning offerings

Amazon, Microsoft and Google provide a number of other MLaaS microservices. Here are a few of note:

  • Release Date: April 2015
  • Description: Amazon Machine Learning is a service that makes it easy for developers of all skill levels to use machine learning technology. It provides visualization tools and wizards that guide you through the process of creating machine learning models without having to learn complex ML algorithms and technology.”
  • Release Date: May 2017
  • Description: Batch AI is a managed service that enables data scientists and AI researchers to train AI and other machine learning models on clusters of Azure virtual machines, including VMs with GPU support. You describe the requirements of your job, where to find the inputs and store the outputs, and Batch AI handles the rest.”
  • Release Date: September 2017
  • Description: Cloud Job Discovery is part of Google for Jobs – a Google-wide commitment to help people find jobs more easily. Job Discovery provides plug and play access to Google’s search and machine learning capabilities, enabling the entire recruiting ecosystem — company career sites, job boards, applicant tracking systems, and staffing agencies to improve job site engagement and candidate conversion.”

Conclusion

MLaaS will be a driver behind AI adoption in 2018, because it makes it simpler for businesses and developers to take advantage of machine learning capabilities. It will fuel the rise of embedded AI in business software applications, and it will allow organizations to use data in new ways that would be otherwise impossible without hiring a highly skilled AI developer.

The lack of machine learning talent in general will be a reason why businesses choose machine learning services from the likes of AWS, Microsoft Azure, and Google Cloud Platform. With these microservices, it is easy to set up and run machine learning algorithms that enhance business processes and operations, customer interactions and overall business strategy. Companies will most likely begin using the services based on other digital platform offerings that they already use for the likes of cloud computing and IaaS. Regardless, MLaaS will become a much more sought after service in 2018.


rob-light

Rob Light

Senior Research Specialist, G2 Crowd