How else could you analyze 36,000 naked mole rat chirps to find out what they’re talking about?
Or translate your cat’s purr or meow to know it’s “just chilling”?
Or auto-generate an image like this just by typing in the words: “giant squid assembling Ikea furniture”?
Thanks to different types of machine learning, that’s all seemingly possible.
4 Types of Machine Learning
- Supervised learning
- Unsupervised learning
- Semi-supervised learning
- Reinforcement learning
What Is Machine Learning?
Machine learning is a branch of artificial intelligence where algorithms identify patterns in data, which are then used to make accurate predictions or complete a given task, like filtering spam emails. The process, which relies on algorithms and statistical models to identify patterns in data, doesn’t require consistent, or explicit, programming. It’s then further optimized through trial and error and feedback, meaning machines learn by experience and increased exposure to data, much the same way humans do.
Today, machine learning is a popular tool used in a range of industries, from banking and insurance — where it’s used to detect fraud — to healthcare, retail marketing and trend forecasting in housing and other markets.
4 Types of Machine Learning (With Examples)
Supervised learning is machine learning with a human touch.
With supervised learning, tagged input and output data is constantly fed and re-fed into human-trained systems that offer real-time guidance, with predictions increasing in accuracy after each new data set is fed into the system. One of the most popular forms of machine learning, supervised learning requires a significant amount of human intervention because of its use of labeled data sets. Data must be divided into features (the input data) and labels (the output data). Human intervention is also needed to provide feedback on the accuracy of the machine learning algorithm, which helps it to learn over time.
Supervised learning, like each of these machine learning types, serves as an umbrella for specific algorithms and statistical models. Here are a few that fall under supervised learning.
Used to further categorize data — think pesky spam and unrelenting marketing emails — classification algorithms are a great tool to sort, and even hide, that data. (If you use a Gmail or any large email client, you may notice that some emails are automatically redirected to a spam or promotions folder, essentially hiding those emails from view.)
Under the broad umbrella of classification algorithms, there’s an even narrower subset of specific machine learning algorithms like naive Bayes classifier algorithms, support vector machine algorithms, decision trees and random forest models that are used to sort data.
When it comes to forecasting trends, like home prices in the housing market, regression algorithms are popular tools. These algorithms identify relationships between outcomes and other independent variables to make accurate predictions. Linear regression algorithms are the most widely used, but other commonly used regression algorithms include logistic regressions, ridge regressions and lasso regressions.
With unsupervised learning, raw data that’s neither labeled nor tagged is processed by the system, meaning less legwork for humans.
Unsupervised learning algorithms work by identifying patterns within a data set, grouping information based on similarities and differences, which is helpful when you’re not sure what to look for. Unsupervised learning is especially useful in customer and audience segmentation, as well as identifying patterns in recorded audio and image data.
Here’s one example of an unsupervised learning algorithm.
Clustering algorithms are the most widely used example of unsupervised machine learning. These algorithms focus on similarities within raw data, and then groups that information accordingly. More simply, these algorithms provide structure to raw data. Clustering algorithms are often used with marketing data to garner customer (or potential customer) insights, as well as for fraud detection. Some clustering algorithms include KNN clustering, principal component analysis, hierarchical clustering and k-means clustering.
Semi-supervised learning offers a balanced mix of both supervised and unsupervised learning. With semi-supervised learning, a hybrid approach is taken as small amounts of tagged data are processed alongside larger chunks of raw data. This strategy essentially gives algorithms a head start when it comes to identifying relevant patterns and making accurate predictions when compared with unsupervised learning algorithms, without the time, effort and cost associated with more labor-intensive supervised learning algorithms.
Semi-supervised learning is typically used in applications ranging from fraud detection to speech recognition as well as text document classification. Because semi-supervised learning uses labeled data and unlabeled data, it often relies on modified unsupervised and unsupervised algorithms trained for both data types.
With reinforcement learning, AI-powered computer software programs outfitted with sensors, commonly referred to as intelligent agents, respond to their surrounding environment — think simulations, computer games and the real world — to make decisions independently that achieve a desired outcome. By perceiving and interacting with their environment, intelligent agents learn through trial and error, ultimately reaching optimal proficiency through positive reinforcement, or rewards, during the learning process. Reinforcement learning is often used in robotics, helping robots acquire specific skills and behaviors.
These are some of the algorithms that fall under reinforcement learning.
Q-learning is a reinforcement learning algorithm that does not require a model of the intelligent agent’s environment. Q-learning algorithms calculate the value of actions based on rewards resulting from those actions to improve outcomes and behaviors.
Deep Reinforcement Learning
Used in the development of self-driving cars, video games and robots, deep reinforcement learning combines deep learning — machine learning based on artificial neural networks — with reinforcement learning where actions, or responses to the artificial neural network’s environment, are either rewarded or punished. With deep reinforcement learning, vast amounts of data and increased computing power are required.