What is Supervised Learning?
Supervised learning consists of inferring a function by mapping an input to an output, based on an input-output set of data. However, this relationship doesn’t imply a cause-effect relationship but merely shows how the two sets correlate with each other. Two typical supervised learning methods are regression and classification.
What is Unsupervised Learning?
In general, unsupervised methods are used to find patterns within a dataset of unlabeled data. For this, algorithms search for data similarity based on underlying hidden features. Data points that are “similar” are grouped together. Two typical unsupervised learning methods are clustering and association.
Semi-supervised learning combines both approaches and uses datasets where only a small portion is tagged data.
Supervised versus Unsupervised Learning
The table below explains the main differences between both types of learning algorithms:
| Supervised Learning | Unsupervised Learning | |
| Training data | Uses input and output values in order to find their relationship. | Uses unlabeled input data in order to find patterns in it. It does not use output data for training. |
| Typical methods | Support vector machine, neural network, linear and logistics regression, random forest, and classification trees. | Cluster algorithms, K-means, hierarchical clustering. |
| Computational complexity | Usually simpler than unsupervised algorithms. | Complex. |
Why are Supervised and Unsupervised Learning Important?
Supervised and unsupervised learning form the core of machine learning. Together they solve important problems in science and engineering. Their importance grew in the latest years, with the advent of big data and more powerful computers.
Both methods are applied to many problems in areas such as marketing, medical science, physics, chemistry, anthropology, economics, finance, etc.
Supervised and Unsupervised Learning + LogicPlum
LogicPlum’s platform allows its users to create models based on both, supervised and unsupervised techniques. It does so by automating the selection, training, and evaluation of different models; and then selecting the optimal one by comparing their efficiencies according to a selected metric.
The advantage of this approach resides in the fact that users don’t need to be experts in machine learning algorithms, but only be capable of understanding the results obtained through the model.
Guide to Further Reading
Wikipedia. Supervised Learning. Available at https://en.wikipedia.org/wiki/Supervised_learning.
Wikipedia. Unsupervised Learning. Available at https://en.wikipedia.org/wiki/Unsupervised_learning.
