by Carol Vercellino, CEO of Oak City Labs
Machine learning is one of the most significant technological advancements in recent history. It’s experiencing incredible growth, and innovators are making breakthroughs in ways that no one ever could have expected.
Because of this, machine learning has changed and will continue to change a lot over the coming years. You’re probably familiar with the buzzword, but the fact remains that many people can’t really answer the question of what machine learning is – and what it isn’t.
In this article (and video), we’ll talk about what machine learning really means, how it’s changed over the years, and where we think machine learning will go next.
What is machine learning?
Machine learning is a process where you use algorithms to parse data, learn from it, and make a prediction.
For example, one of the most common ways people are using machine learning right now is with stock market predictions. You can try to predict what a particular stock is going to do by combining historical stock data with other data, like weather, GDP, and even election data.
On a more personal note, if you have an iPhone, you can label your pet, child, partner, or friend in one of your photos. You’ll then notice that Apple Photos goes back through your album and labels any other pictures of that person or animal. You can also use search terms like ‘cake’ or ‘beach’ to sort through photos. That’s machine learning on the back end.
Machine learning is not where you have a whole bunch of data that people – or statisticians – manually go through (to learn more about the differences between machine learning and statistics, check out this article).
It’s an automated process, and the data is like a black box to the machine. The machine takes that data, dumps it into different models, and tries to develop the best possible answer. So, the machine doesn’t care whether it’s exploring weather or GDP data.
A brief history of machine learning:
How does machine learning work?
Within machine learning, there are 3 different types of learning:
- Supervised learning
- Unsupervised learning
- Reinforcement learning
Supervised learning is when you give labeled data to the machine. For example, you have a whole data set full of cats and dogs. You give the machine that data set, and the machine learns what a cat is versus a dog. Then, you give the machine a photo that hasn’t been labeled and ask the machine to identify it as a cat or dog. And the machine can do that using algorithms and a little bit of human intervention.
Unsupervised learning is when the data is unlabeled. Take the same example, but instead of giving the machine images labeled as cats and dogs, you leave the photos unlabeled and ask the machine to look for patterns in those images. Then, you can come up with a label for those patterns.
Reinforcement learning is much like playing a video game. In a video game, the user goes through a level and gets a reward or badge before proceeding to the next level. Reinforcement learning is very much like that. The automation or algorithms attempt to find the most optimal way to accomplish a task.
When the machine does well, it moves onto the next task, and it continues to try and get better and better each time.
What’s next in machine learning?
Machine learning is becoming more commonplace, but we’re also moving towards getting more information around the data.
According to Jason Burke of CREO, Inc., the biggest problem in the machine learning space right now is not the algorithms or math; it’s cleaning the data and the context of the data. We need to combine statistics and machine learning to get more context around the data itself.
We believe we’ll see advancements in that area in the coming years, plus more user-friendly platforms where anybody can sit down and use data science and machine learning.
Machine learning will undoubtedly continue to disrupt every industry. How are you going to use machine learning to impact your company, industry, or the world?
Speaking of advancements in technology…here’s what you need to know about Apple’s new iPads, watches, and the iOS 14 release.