You should always love your grandparents. They give great advice, have answers to every question, are always up for an adventure, and give you more money than your parents do. Grandparents are amazing — except when you’re trying to show them how to use their iPhone.
Humans learn from past experiences.
Machines follow instructions given by humans.
What if humans can teach a machine to learn from the machine’s past experiences? That is machine learning.
What is machine learning? Our ability to learn and get better at task through experience is part of being human. When we’re born we know little to nothing and can do little to nothing for ourselves. As time goes on, we become more capable and are able to do more things everyday. Computers can grow in the same way. Machines can start from little to nothing and, through experience, can learn to become more capable and can improve as time goes on.
Machine Learning is Romantic
This is a computer. His name is Tarzan. He grew up on a deserted island, so he’s never seen a woman in his life. Jane arrives on the island. He gives her a rock, and she doesn’t like that. He’s never dated, so he doesn’t know what women like. So Tarzan keeps on trying. He collects data. She seems displeased when he yells, but she likes it when he smiles. This is data. She doesn’t like it when he brings her dead snakes to eat, but she likes coconuts. More data. While building such experiences, he learns how to win Jane’s heart. Now Tarzan knows a few things. He knows she likes coconuts and smiles, but dislikes dead snakes and yelling. While looking for food, Tarzan stumbles upon two things. A dead rabbit and a mango. Tarzan does not have any data on, or experiences with, dead rabbits and mangoes. There are two things our computer, Tarzan, can do.
One, he can try giving it to Jane and see if she likes it.
Or two, he can make a prediction.
She doesn’t like dead snakes and likes coconuts. The dead snake is similar to a dead rabbit so there’s a strong chance she won’t like it. The mango is similar to a coconut so there is a possibility she will like it. If he has to choose one, he will probably go with the mango. Tarzan will remember Jane’s reaction to the mango and will use the result at a later time to make a different decision.
THE BASICS
Machine learning is a field of study that gives computers the ability to learn without being explicitly programmed. It is a very exciting and growing field that allows computers to be have more like humans. It is a lot different from how most developers program traditional machines. In traditional computing, very clear instructions are given to the computer. Computers are really dumb and have to be told exactly what to do and how to do it. With machine learning we can give a computer lots of data to analyze and then that machine can use that data to learn. There are many ways a machine can learn. I want to talk about each one. The three ways of machine learning are supervised learning, unsupervised learning, and reinforced learning.
Supervised learning
Suppose your friend gives you one million coins consisting of pennies, dimes, and quarters. Each coin has a different weight. If you tell the computer what each coin is, the computer can notice that each coin has a different weight and a different size. The machine will learn that a penny is the smallest and the quarter is the largest. This is supervised learning because you are already telling the computer “this coin is a dime, this coin is a quarter”. There is labeled data.
Unsupervised learning
Let’s use the same coin example, but this time the computer doesn’t know the names of the coins. Instead you ask the machine to weigh the coins and you tell the machine that quarters are the largest and pennies are the smallest. Will the machine be able to tell which coin is which? This is known as unsupervised learning because you are asking the computer to label things on its own. There is no labeled ata.
Reinforcement learning
Reinforcement learning is a reward based learning. It works by giving feedback to the computer; either positive, or negative. Here, let’s say you give a computer a picture of a dog and ask it to tell you what it is. If the computer says its a cat, you provide negative feedback. If the computer comes across the same image, it will know that it is not a cat. It comes across a similar image, it will remember the previous image and use the experience to make a decision.
Grandpa, why is everyone talking about machine learning? It’s becoming increasingly useful and relevant for three reasons. First, there’s more data than ever before. Second, computers are more powerful than they used to be. Third, there are better machine learning techniques. Machine learning surrounds us. It’s used to recommend purchases on the internet, it determines what ads you see on your phone, and it can even detect fraud on your credit card. Let me give you some examples of how machine learning works when they have a lot of data, or experiences, to look at.
When you use Netflix, you may notice that the website recommends things for you to watch. How does it do that? First, Netflix pays attention to what you watch and how you react to it. If you watch the Lion King, Frozen, and Moana and say that you like it, Netflix will look at other users who also watched those movies and liked it. If 1000 other people watched those same three movies and liked it, then Netflix can recommend other movies those 1000 people watched and liked. There is evidence that you will like the same movies that they did, too. This is called the nearest neighbor algorithm. The machine matches your interests with the interests of people similar to you.
The Future is Machine
The rise of more and more machine learning is inevitable. Today it already plays a leading role in most popular technologies. Spotify uses machine learning to create daily mixes for you based on what you and people like you are listening to. Amazon uses machine learning to recommend products that you are more likely to buy.
Businesses are obsessed with machine learning because it can automate tasks typically reserved for humans. A lot of companies, like Amazon, use chatbots for their customer service. The next time you are chatting with an “employee”, there might not be an actual human behind the other end of the computer.
Machine learning is useful but can create disastrous consequences for the future. Machine learning is being developed for use in autonomous cars and trucks. This can make truck drivers obsolete and take away jobs from over 3 million Americans. Driving trucks is one of the largest occupations in the United States. What will happen when companies no longer have to hire a driver to move their product across the country?
As machine learning continues to be embedded in every aspect of technology it is important to not let the positive business benefits distract society from the negative consequences of relying too heavily on machines.
[1]: Daniel Faggela. (November 21, 2019). What is machine learning?https://emerj.com/ai-glossary-terms/what-is-machine-learning/
[2]: Jason Brownlee. (August 16, 2019). What is deep learning?https://machinelearningmastery.com/what-is-deep-learning/
[3]: Atul. (May 22, 2019). What is Machine Learning? Machine Learning For Beginners https://www.edureka.co/blog/what-is-machine-learning/
[4]: Royal Society. (May 22, 2017). Machine-Learning
https://royalsociety.org/topics-policy/projects/machine-learning/videos-and-background-information/