Machine learning is a branch of Artificial Intelligence (AI), and as defined by the American computer scientist and E. Fredkin University Professor at the Carnegie Mellon University (CMU), Tom M. Mitchell, “Machine learning is the study of computer algorithms that allow computer programs to automatically improve through experience.” Major search engines like Google and Bing use it to provide the best links, and entertainment websites like Youtube and Netflix use it to give recommendations. It’s also used in speech and facial recognition, self-driving cars, targeted ads, and much more. Machine learning is similar to that of human learning - it depends on experience and trial and error.
Imagine a task like organizing your laundry. You want to be able to organize and classify clothes so that it is easier for you to find it later. To do something like this, your brain has to find a pattern in these garments - how they are shaped, how you wear them, what they are made of, and more. For you, these patterns have existed in your brain for a long time. Your brain then uses these patterns to identify what goes together: t-shirts in one pile and pants in another. Now imagine a computer algorithm doing something similar, like organizing photos of cats and dogs. To start, the computer algorithm is given pictures, or data, and tries to find a pattern in them - certain features dogs have that cats don’t. They then can apply these patterns, and now when you give the computer algorithm a dog picture, it will be able to identify that it is a dog and store it away in a dog picture database. Of course, it's much more complicated than that, but this is the basis of machine learning.
A model of brain cell interaction created by Donald Hebb in 1949 in a book titled The Organization of Behavior presented his ideas of communication between neurons. Those concepts could be applied to artificial neural networks and artificial neurons (nodes). In the 1950s, Arthur Lee Samuel, a pioneer in computer gaming and artificial intelligence made a computer program for playing checkers. He added features that allowed the program to learn from experience, and thus the term “machine learning” was created. Frank Rosenblatt went on to use both these creations to create software for the Mark 1 Perceptron which was used for image recognition. It failed to do so for many visual patterns, and research for machine learning died down. However, in the 1960s, both basic pattern recognition and neural network research received a kickstart with the nearest neighbor algorithm and multilayers.
In 1986, Geoffrey Hinton showed that backpropagation could solve what went wrong with the Perceptron; it could train a deep neural network, which was what started deep learning.
A subset of machine learning, deep learning is a technique that gives machines the ability to find even the smallest patterns and enhance them. It’s extremely powerful due to the many layers of artificial neural networks stacked on top of each other that filter through the staggeringly huge amount of data we have.
Machine learning and deep learning come in three main categories, each with their own applications. Supervised learning labels the data so the algorithm knows exactly what to look for - it’s what powers your YouTube recommendation page. Unsupervised learning is less prevalent. The data isn’t labeled and the algorithm just looks for patterns. Reinforcement learning, as the name suggests, rewards or penalizes the machine while it tries to reach an objective through trial and error.
Machine learning is much more complicated than this of course, but the most important thing you have to remember is that it’s how machines take in data, find a pattern, and apply the pattern.
Copeland, B.J. “Artificial Intelligence.” Encyclopædia Britannica, Encyclopædia Britannica, 11 Aug. 2020, www.britannica.com/technology/artificial-intelligence/Is-strong-AI-possible
Foote, Keith D. “A Brief History of Machine Learning.” DATAVERSITY, DATAVERSITY, 13 Mar. 2019, www.dataversity.net/a-brief-history-of-machine-learning/
Hao, Karen. “What Is Machine Learning?” MIT Technology Review, MIT , 2 Apr. 2020, www.technologyreview.com/2018/11/17/103781/what-is-machine-learning-we-drew-you-another-flowchart/
Marr, Bernard. “What Is Deep Learning AI? A Simple Guide With 8 Practical Examples.” Forbes, Forbes Magazine, 12 Dec. 2018, www.forbes.com/sites/bernardmarr/2018/10/01/what-is-deep-learning-ai-a-simple-guide-with-8-practical-examples/
Somers, James. “Is AI Riding a One-Trick Pony?” MIT Technology Review, MIT, 2 Apr. 2020, www.technologyreview.com/2017/09/29/67852/is-ai-riding-a-one-trick-pony/