Google Cloud AI Services Deep Dive
Google Cloud Training Architect II in Content
Welcome to the Google Cloud AI Services Deep Dive course. I'm Joe Lowery, and I'll be your guide through this amazing collection of technology that will, without a doubt leave you wide-eyed in wonder at the possibilities of what you can accomplish at this very moment.
Artificial intelligence in general and machine learning in particular have increased the reach and capabilities of technology exponentially, completely altering our world at an amazing pace. In this course, we'll explore a large number of the Google Cloud services — 15 in all — that are available for you to incorporate into your own applications and organizations.
First off, we'll provide some context for the overall course and describe the differences between artificial intelligence, machine learning, and deep learning. Then, we'll look at one of the key components of the Google Cloud AI services we'll be examining, AutoML, as we detail its unique features.
Then, we'll dive into specific target categories with a look at the AI Sight services, including those that can detect objects or text in still images or even video.
From imagery, we move to AI services for written language with tools that can extract the structure and meaning of text — including sentiment — and ones that can translate from one language to another, in over 100 different varieties.
The AI Conversation services provide the power to move from one form of communication — speech — to another — text — and back again. In addition, you'll see how Google has brought many of their tools together to create a service that can build conversational interfaces to interact with your users through almost any medium.
In our final section, we'll explore AI services dedicated to another, omnipresent method of communication: data. Within the AI Structured Data group, you'll find a tool for analyzing and making predictions with tabular data, one for recommending products based on user events, and another that can create machine learning models with standard SQL queries.
Obviously, there's a tremendous amount of power and complexity involved in Google Cloud's AI services. Understanding their potential will increase your own for developing the applications, systems, and organizations of the future.
Google Cloud AI Services Deep Dive
Welcome to the Google Cloud AI Services Deep Dive course. I'm Joe Lowery, and I'll be your guide through this amazing collection of technology that will, without a doubt, leave you wide-eyed in wonder at the possibilities of what you can accomplish at this very moment. Artificial intelligence in general and machine learning in particular have increased the reach and capabilities of technology exponentially, completely altering our world at an amazing pace. In this course, we'll explore a large number of the Google Cloud services — 15 in all — that are available for you to incorporate into your own applications and organizations. In this lesson, you'll get a 10,000-foot view of the services and subjects covered.
About the Training Architect
Meet the training architect of this course, Joseph Lowery. Joe has been working with Google Cloud for over five years, transitioning websites to the cloud via App Engine, Compute Engine, Cloud Storage, Cloud Datastore, and other services. He is Linux Academy's training architect for Google Cloud Essentials, Google Kubernetes Engine Deep Dive, Google App Engine Deep Dive, Google Cloud Functions Deep Dive, and Google Cloud Apigee Certified API Engineer as well as a full slate of hands-on labs for Google Cloud.
What Is AI/ML?
Chances are, the concept of artificial intelligence has been around not only as long as you've been alive but probably for most, if not all, of your parents' lives. Yet, the majority of us — unless you've specifically studied the field — don't have a clear idea of what is meant by the term artificial intelligence. Furthermore, the phrase machine learning has been gaining popularity recently — again, without a broad understanding. In this lesson, we'll define exactly what we mean by both these terms in order for us to begin building a foundation for the following lessons in this course. In addition, we'll bring a third one into the mix: deep learning — a necessary model for understanding the state of Google Cloud AI services today.
Understanding Google Cloud AI and Machine Learning
Google Cloud's offerings in the AI field are extensive, comprehensive, and — quite frankly — somewhat overwhelming. In this lesson, we'll take a look at the full range of what's available, from the underlying AI Platform, to the plug-and-play components of the AI Hub. There's a lot to cover, so let's get started. Link referenced in lesson: https://aihub.cloud.google.com/
Targeting Cloud AutoML
In the previous two lessons, you've seen what a far-reaching goal artificial intelligence has as well as how complex machine language and deep learning can be. You've also seen how complex establishing a trained machine learning model, complete with self-modifying predictions, can be. In this lesson, we cover how Google Cloud makes all of that approachable so that developers like yourself can add machine learning functionality to their applications in selected fields such as sight, language, and structured data.
Identifying Images with Vision AI
Since the advent of computers, digital imagery has exploded — both in the sheer number of photos available and in their societal impact. The Google Cloud AI services related to vision — such as AutoML Vision and the Vision API — have turned this overwhelming avalanche into an opportunity for advancement and automation, whether it's used in industry to identify defects of windmills, in apps to pinpoint similar products, or online to restrict certain types of content. In this lesson, we'll take a look at all of Google Cloud's various vision-related AI services and put AutoML Vision through its paces.
Examining Video AI
In the previous lesson, you saw how Google Cloud AutoML Vision could easily detect figures in a single image. That takes a fair amount of computing power. Imagine how much more it would take to identify objects in a digitized video, with frame rates up to 60 frames per second. But that's exactly what the Google Cloud's AI video services are capable of — and more. In this lesson, we'll cover the capabilities and uses of both AutoML Video Intelligence and the Video Intelligence API.
Extracting Data with Natural Language
The previous section dealt with AI and imagery, both still and moving. But what about our other primary method of communication: text? Google Cloud has a solid slate of services dedicated to analyzing and dissecting all manner of documents: text files, invoices, receipts, PDFs, emails, social media, and much more. In this lesson, we'll examine the two machine learning services that target natural language, detailing their capabilities as well as their differences, and walk through a demonstration of what's possible.
Google Translate is one of the earliest and best-known examples of Google technology. With this browser- and app-available service, users can translate phrases and even whole documents between — at last count — 103 languages. This incredibly valuable service is powered by Google's own neural network-driven Translate API. As you'll discover in this lesson, Google actually offers a range of translation services that not only give access to the core Translation API, but also allow custom language terms and phrases to be integrated into a trained machine learning model.
There's a scene in one of the older, classic Star Trek films — Star Trek IV: The Voyage Home — where the team has been transported back in time to 1986. Scotty, the lead engineer, walks confidently up to a police detective's Macintosh, picks up the mouse like a microphone, and says, "Computer… Computer… hello, Computer." with zero results. Today's engineers — even today's kids — think nothing of calling out to Siri, Alexa, or Google to submit a request, made possible by advancements in artificial intelligence's speech-to-text capabilities. In this lesson, we'll examine Google Cloud's Speech-to-Text offerings and detail exactly what their capabilities are as well as typical use cases. We'll close out the lesson with a brief demonstration of Cloud Speech-to-Text in action.
In the previous lesson, we saw how Google Cloud makes it possible to transcribe recordings of people speaking — now it's time to reverse the process and turn text into speech. The movement to output synthetic speech is more than just a challenge — it's a path toward integrating computing into our lives. Typical use cases include call center automation, responses from IoT devices, and converting text to audio — all of which enable machines to interact with people in a more natural manner. In this lesson, we'll examine Google Cloud Text-to-Speech and describe its capabilities and options.
Conversing with Dialogflow
So far, we've discussed a number of speech-related AI services, including Cloud Natural Language, Cloud Speech-to-Text, and Cloud Text-to-Speech. In this lesson, we'll cover a service that brings all of those (and more) together: Google Cloud Dialogflow. The goal of Dialogflow is to bring vocal conversations between a human and an AI service to computing applications. This is understandably a massive objective, as conversations can turn on a dime and are filled with many non-standard expressions. Remarkably, Google is able to achieve great success with their efforts — so much so that people engaging with Dialogflow may not be aware they're talking to an AI service. And even if they do realize it, the achieved efficiency is well worth the cost. As part of this lesson, we'll take a look at the steps for creating a key element in Dialogflow, the agent, in the dedicated UI.
AI Structured Data
Structuring AutoML Tables
The world runs on structured data. Structured data — often referred to as tabular data — is found in every business, whether they are in the financial or consumer sector. This data accumulates quickly and, while it may not be particularly deep, it can be extremely broad and can — when properly analyzed — be used to predict future behavior and actions. In this lesson, we'll examine how Google Cloud AutoML Tables turns tabular data into actionable predictive insights in a massively increased efficiency and scale.
I remember when I first encountered a recommendation engine's results. When Amazon first came on the scene, all they sold were books — and I was — and still am — a voracious reader. After selecting my first book for purchase, I was presented with a list of five other books I might be interested in — and, sure enough, I was. Recommendation engines have become much more sophisticated since then and, in this lesson, we'll get up to speed on one of the latest iterations: Google Cloud Recommendation AI, which is built with the knowledge gained from years of delivering recommended content to properties such as Google Search, Google Ads, and YouTube.
Executing BigQuery ML
When it comes to analyzing and working with structured data, Google Cloud has one of the world's most respected and powerful engines: BigQuery. So, it's only natural that the service be extended to work with machine learning models. This brings the power of machine learning to the vast number of practitioners who are SQL savvy — as all interactions with BigQuery ML are handled with standard SQL queries. In this lesson, we'll take a quick tour of the service, highlighting its advantages and capabilities, and then we'll see its power and simplicity by querying a sample Google Analytics dataset.
Congratulations on completing Google Cloud AI Services Deep Dive! Let's take this lesson to revisit the material and review a few key points.
Now that you've completed this course, here are a few ideas to consider regarding what you should do next.
Take this course and learn a new skill today.
Transform your learning with our all access plan.Start 7-Day Free Trial