What is Machine Learning?
As we have come to learn, artificial intelligence and machine learning, while often used interchangeably, are not one in the same. Artificial intelligence is the overarching term, referring to the science behind the engineering of the computer learning process, while machine learning is a subcategory in which the coding improves itself over time with no outside interference. Machine learning is an ever-growing field and many of the services we use today such as Spotify, Netflix, and YouTube are powered by machine learning.
Everything from what links you are clicking on, to the genres of music you listen to, to the television you watch is being collected by these services. This data allows determination, and prediction of, your interests to ultimately provide a better service for you. While this “finding a pattern, applying the pattern” process may seem basic, it is essentially running the world.1
Deep learning and neural networks are also subcategories of artificial intelligence, but deep learning is a subcategory of machine learning, while neural networks are a subcategory of deep learning. Deep learning focuses on less human intervention than even machine learning requires and can take unlabeled datasets and automatically determine distinguishing features therefore creating its own categories.2 Neural networks are node layers that send data to the next layer of the network. If the neural network has more than three layers, that is what makes it a “deep learning” algorithm.2
There are three kinds of machine learning: supervised learning, unsupervised learning and reinforcement learning, and they are pretty much what they sound like. Supervised learning uses prelabeled datasets for accurate predictions based on the classifying data algorithm.1 With unsupervised learning, the data is provided unlabeled requiring the machine to find patterns on its own.1 Reinforcement learning is a trial-and-error method in which the machine tries different techniques to complete its objective. IBM’s Watson is an example of reinforcement learning in that it was rewarded for making good decisions when it came to how much to wager and which squares to play when it won Jeopardy in 2011. If Watson had played incorrectly, it would have been penalized by the game moving on to the next player.
While it all sounds kind of similar, it is important to understand the vast differences between all the types of learnings under the umbrella of artificial intelligence. It is more than just getting a camera to identify people with a weapon, or one day creating a robot with emotions. It is countless hours of developing datasets, algorithms, and trial runs to get the desired outcome for your AI creation.
Hao, Karen. “What Is Machine Learning?” MIT Technology Review, MIT Technology Review, 5 Apr. 2021, https://www.technologyreview.com/2018/11/17/103781/what-is-machine-learning-we-drew-you-another-flowchart/.
By: IBM Cloud Education. “What Is Machine Learning?” IBM, 15 July 2020, https://www.ibm.com/cloud/learn/machine-learning.
