Machine Learning for Dummies (like me)

There is a lot of conversation happening around how artificial intelligence, and specifically machine learning, will be the next great disruptor to industry. Some are going as far as to call it the new "electricity". Similar to rhetoric about the cloud years ago, many people are talking about AI but relatively few actually understand it. The truth is, AI concepts and technologies can be intimidating; they were intimidating to me and I've been an emerging technology engineer for over 20 years. But I wanted to have more meaningful conversations about the topic with our data scientists and customers, so I've dug in, shored up my understanding and have summarized my learnings into a primer below. It's not short but I've tried to make it meaningful, comprehensive and hopefully, easy to follow.

One thing I can assure you of is you don't have to be intimidated by these technologies. The concepts are not difficult to grasp; but the potential for these technologies is undeniable. I do believe they will fundamentally change how businesses operate. AI and machine learning have the potential to drive significant cost out of operations while creating never before possible customer experiences. This primer is meant for a non-engineering audience, in hopes that all may benefit from understanding and embracing the opportunities these technologies will provide.

What is Machine Learning?

Machine learning is defined as the science of getting a computer to act without being explicitly programmed. But what exactly does that mean? The 'machine' part of machine learning can be anything. It could be an app, a car, a robot, a server in a data center or an API in the cloud. The 'learning' piece refers to a program that gets smarter as more data is provided to it. In fact, machine learning works in a similar fashion to how our brain learns. Let's look at an example.

Much of our brain works on the concept of pattern recognition. For example, our brains use pattern recognition to translate the sounds someone is speaking into words we can understand. For a moment, think about this process of language comprehension as a sound-to-word-translation-program. Someone says "hey", your sound-to-word-translation-program runs and you respond back with, "hey". If the person we're talking to is someone we've never met before, and therefore we've never heard their exact sound for "hey", we can still understand them. We are able to do this because our mind has created a massive collection of all the different sounds for "hey" we have previously heard and has indexed them into a word group of "heys". So if we meet someone new and they say, "hey", we scan all of our word groups for that sound until we find the one group that provides the closest match. This is pattern recognition. 

One interesting thing to note is our sound-to-word-translation-program gets better over time. For example, when an American meets someone from Scotland for the first time, it may be harder for them to understand them at first. But once they do, they get better at understanding everyone with a Scottish accent. How does that work? If our brain hears a sound we haven't heard before (i.e. can't find a word group) we say "sorry, I didn't understand you, can you say that again?". We then listen more carefully for the new sound and scan our existing word groups again. When we do end up finding the right word group, we file that new sound into that group, making us even better at recognizing that word (or that accent) in the future. We're now smarter at recognizing this new sound, but we haven't reprogrammed our mind. We've just added new data to our existing sound-to-word-translation-program. 

At a high level this is how machine learning works. We write the program once, but the program gets better over time as we add more data to it. The more data we give it, the smarter it gets. Amazon's Alexa is a great example of machine learning. Most people's sound-to-word-translation-program is pretty good and operates at about 95% accuracy. Amazon's Alexa uses machine learning to run a similar program and her accuracy is almost 94%. But Alexa is getting smarter the more people use her. Right now she is pretty good at understanding my wife and I, but she is not very good at understanding our Ukranian nanny's accent. Alexa's also not very good at understanding my 4 year old, since her word pronunciation is still developing. But over time, Alexa will get better as she collects more data from these demographics. And she's not just collecting this data in our house, but in everyone's house that's using her. Eventually, she'll be better than any human on the planet at understanding people with various accents and dialects across many different languages; who knows, she might even be able to understand dolphins. 

This is why machine learning is often used interchangeably with artificial intelligence. With machine learning we can simulate how our own brain works to allow machines to do things that historically, only humans could do. Machine learning powers things like the natural language processing in Alexa and Siri, and it also powers things like self-driving cars. Self-driving cars, are using visual pattern recognition of highway lanes, curbs, other cars and street signs (along with many other sensor data points) to understand their surroundings and make decisions on when to hit the gas, when to hit the brake and when to turn the wheel. Similar to a 16 year-old who just got her license, the more a self-driving car drives, the better at driving it gets, because it has more patterns and experiences (data) to draw from. The advantage self-driving cars have over the 16 year-old however, is that every self-driving car from a manufacturer can share one massive dataset of patterns or experiences to draw from. This is why many people believe that, over time, self-driving cars will be much better at driving than humans. 

Machine learning can also be used to do things humans can't do. For example, machine learning is used to return Google search results instantaneously, sequence the human genome, predict traffic times and spot fraudulent activity on your credit card. You are already interacting with machine learning dozens of times every day without knowing it. But even still, the applications of machine learning are extremely limited compared to the potential use cases, which is why many feel it will be a big disruptor as it continues to be applied to different industries. 

What Are the Applications of Machine Learning in Business?

There are two broad categories of machine learning: supervised learning and unsupervised learning. Each have different practical applications.

Supervised Learning

Supervised learning is when we have a clear idea of the output, or type of answer we want based on the input we put in. Every example we've given thus far is an example of supervised learning. We put in a sound, we get back a word. We put in a set of sensor data, we get back a decision of whether to speed up or slow down the car. We put in square footage, bedrooms and location, we get back a price. 

Supervised learning can further be broken down into regression and classification models. Regression machine learning models follow some continuous output. For example, the bigger a house is, the more its price will probably go up. Or the more traffic there is on 290, the longer my commute will probably take. Regression machine learning models generally follow what one would think of as a "trend line" (for you Excel users). But think of a trend line that isn't just two dimensions (i.e. sq footage as a function of price) but one that could be 3 or 4 or 20 dimensions (depending on how many features we want to include). That's regression machine learning. 

For applications of regression machine learning, think about the moments that matter in your business or department that you would like to more accurately predict; or even better, proactively address. Inventory shortages? Cash flow? Customer wait time? Returns? Maintenance tickets? The close date of a big sale? Chances are these outcomes are a result of past actions or external factors (like weather, traffic or economic indicators) or a combination of the two. Regression could be used to predict the timing or magnitude of these moments that matter.  

Classification machine learning models on the other hand are about putting things into buckets or categories. For example, if you give me an image of a tumor, I'll tell you if it's malignant or benign; if you give me an email I can tell you whether or not it's spam. In GMail, classification is further used to bucket an email as Important, Promotional or Social.

Opportunities to apply classification machine learning are often seen in the repetitive tasks performed by knowledge workers: Lawyers scanning through contracts looking for specific language, technicians scanning images for certain characteristics, contact center reps going through a call script or decision tree with a customer, process or data analysts reading an email and routing it to a next step in a process. These tasks can often be automated through the application of classification machine learning techniques.

Unsupervised Learning

The other type of machine learning is unsupervised learning. This is where we don't know what the output is and we actually want the machine to give us some suggestions or insights based on correlations it can find in the data that may not be obvious to us on the surface. Unsupervised learning has been used to determine what gene mutations might be responsible for certain medical conditions. Amazon and Netflix utilize unsupervised learning to power their recommendation engines. LinkedIn and Facebook use it to suggest connections for you to make. Similar to supervised learning, the more data these unsupervised learning systems have, the smarter they become. 

Opportunities to apply unsupervised learning techniques exist where correlations in your business could be valuable but you may not know where to look. If you want to ask a more open ended question about your business, unsupervised learning can offer a different perspective. For example, what attributes do all your top customers have in common? Maybe most of them are veterans at their company for more than 4 years, or a disproportionate amount of them are female directors, or are active LinkedIn users. This information may influence your marketing strategy. What's unique about your company's top sales performers? Maybe a majority of them have liberal arts degrees, or speak a second language or are active on a charity board. This could influence your sales recruitment strategy. Unsupervised learning can provide insights that may not be visible on the surface, which might lead to discovering moments that matter that you didn't realize existed.

_____________________________________________________________________

It's been estimated that 85% of the data in existence is within corporate data centers. This provides a massive competitive advantage for many organizations if they choose to take action by identifying the machine learning use cases for their business. Together we can further demystify while accelerating the adoption of this next-great advancement in technology.

-J