Artificial Intelligence. Machine Learning. Deep Learning. You’ve heard the teams over and over again and you are fully aware that these technologies are very much on your company horizon. After all, everyone is making use of at least one of them. But what are they? What’s the difference between them?
You’ll find examples of these everywhere. Netflix, Amazon, Google, Microsoft, and all of the big companies are making use of one form or another. For example, you go to photos.google.com and search for “cats.” Now, you haven’t titled or tagged any of your photos with the label “cats,” and yet somehow the service is able to recognize any photo on your cloud storage that contains a cat.
It’s not magic. It’s AI.
And that’s at the heart of this discussion: Artificial Intelligence. You see, Machine Learning is a subset of AI, and Deep Learning is a subset of Machine Learning. Confused yet?
Let’s first define Artificial Intelligence.
What is AI?
AI is when a machine attempts to simulate human intelligence. Systems that use AI include natural language processing, speech recognition, and machine vision. One of the most prevalent examples of AI is the chatbot, which is an artificial intelligence program that simulates human conversation. Chatbots are widely deployed by businesses as the first line of contact with customers. A chatbot will interact with a customer to either try to resolve an issue or, if it’s incapable of doing so, pass the user on to an actual human.
Chatbots are incredibly useful, efficient, and cost-effective solutions for businesses. Because chatbots can learn based on the input they receive, they are capable of making changes based on patterns. The more interaction they have, the more they learn.
That’s artificial intelligence at its core. In the end, the goal of AI is to make a machine behave in ways that would be called “intelligent” when compared to actual human behavior. In simplest terms, AI is programming machines to behave like humans.
But where does Machine Learning come into play?
What is Machine Learning?
Now that you have a fundamental idea of what Artificial Intelligence is, let’s take a look at Machine Learning. Machine Learning takes AI one step further. Artificial Intelligence requires human input for it to actually learn. The goal of Machine Learning is to make it possible for machines to learn things without having to program those things in the first place.
Machine learning is a subset of AI and is one method of achieving Artificial Intelligence. The goal of Machine Learning is to create a piece of software, feed it data, and then allow it to use that data to learn and improve over time.
One very good example of Machine Learning is recommendation engines on Amazon. Every time you visit Amazon and search for (or purchase) a product, Machine Learning uses the data it receives from your searches and purchases to learn how to better recommend products for you. At first, the recommendations will seem off the mark. Over time, however, the system learns more and more about you and is capable of better recommending products you might actually want to purchase.
That’s Machine Learning.
It’s also important to understand that all Machine Learning is a form of Artificial Intelligence, but not all Artificial Intelligence is a form of Machine Learning. Our chatbot example is not Machine Learning. However, you could design a chatbot with Machine Learning that would be capable of learning in such a way as to be better capable of answering questions. At that point, you’d have a Machine Learning-based chatbot.
The key difference is that the Machine Learning chatbot is capable of actually learning from the input it receives. One of the most popular methods is Natural Language Processing (NLP), which refers to the interaction between computers and human language.
A true Machine Learning chatbot should be able to:
- Offer an informative answer.
- Maintain the context of the dialogue with the interacting human.
- Be indistinguishable from the human.
The biggest hurdle is the final point. This is where the Turing Test comes into play, which is a method of inquiry in Artificial Intelligence to determine whether or not a computer is capable of thinking like a human being.
Eugene Goostman is the only chatbot that some have considered as having passed the Turing Test.
And now we get to Deep Learning.
What is Deep Learning?
Deep Learning is a subset of Machine Learning (which, in turn, is a subset of Artificial Intelligence). Where Machine Learning is accomplished by humans feeding information to a machine, Deep Learning accomplishes the same task through the use of a specific algorithm type called an Artificial Neural Network (ANN).
The biggest difference between Machine Learning and Deep Learning is that in Machine Learning the learning process is supervised. Let’s go back to our cat example. In Machine Learning a developer must be very specific about what things the program should look for to determine if an image contains a cat. Maybe pointy ears, fur, almond-shaped eyes, four legs, and a tail.
With Deep Learning, the program builds this collection of details on its own, and eventually (once it’s gathered all of the necessary bits that can determine if a picture is, in fact, a cat), it can then pick out images of cats in photos.
With Machine Learning, the developer must tell the program how to determine if a cat is in a photo. With Deep Learning, the algorithms (over time) piece together everything it needs to determine if a cat is in a photo. One is supervised, one is not.
One of the more popular languages for Deep Learning is C++.
Conclusion
And there you have it, the difference between Machine Learning and Deep Learning (with a nod to Artificial Intelligence). The most important thing to come away with is that your company will, at some point in the near future, depend on one of these technologies to help expedite your digital transformation and keep up with the competition.
0 Comments