The evolution of AI has changed the entire 21st century. Within AI, machine learning focuses on the development of computer programs that can access data. The essence of machine learning is that algorithms can give computers the ability to learn from data, and make predictions and decisions. In this passage, it will introduce two major methods of machine learning, supervised and unsupervised learning.

Supervised learning can be compared to learning in the presence of a supervisor or a teacher. Differing it from unsupervised learning, supervised learning is provided with labeled data, where the correct answer is given for each input. It is like a function y = f(x), where we already have input variables (x) and output variables (y), and we aim  to learn the mapping function (f(x)) from the input to the output by using specific algorithms. Differing it from unsupervised learning, supervised learning are provided with labeled data, where the correct answer is given for each input. 

Supervised learning models can be further categorized into classification and regression. Classification is the organization of labeled data (e.g. predicting gender of a person), whereas regression is when we want to map the input to a continuous output (e.g. predicting nationality). In both classification and regression, the goal is to find the relationship or structure of the input data that then allows us to produce correct output data effectively. Some considerations including model complexity and the bias-variance tradeoff has to be considered when conducting supervised learning, that a high-complexity model will overfit(fits well to the training data, but does not generalize to other data points) if there’s no enough data points, and increasing bias can result in model with relatively guaranteed baseline level of performance. 

On the other hand, unsupervised learning is to let the model work on its own to discover information that may not be visible to humans. It mainly draws conclusions on unlabeled data. Unsupervised learning algorithms are more complicated and difficult, which allows more complex tasks to be performed although unsupervised learning can be more unpredictable compared with other learning methods including deep learning, reinforcement learning and so on. There is little information about the data that can be provided to the model, including the type of data or the answer of data, which means the algorithms has to look for clusters density estimation, and dimensionality reduction. 

Unsupervised learning includes clustering and association. Clustering is the analysis of patterns and groups of unlabeled data (e.g. k-means clustering), whereas association is where you want to discover rules that describe the majority of the data (e.g. Apriori algorithm). In comparison with supervised learning, unsupervised machine learning has fewer tests, fewer models, resulting in a less controllable environment. 

Difference between Unsupervised and Supervised Learnings

Right now, data scientists are trying their best to comprehend machines but it will all change soon, Falon Fatemi, founder of Node, predicted that unsupervised deep learning is the future. Since there’s a skill shortage in the field of AI, and that teaching a computer is already challenging enough, Fatemi stated that “Unsupervised models can essentially be trained on the knowledge that exists on the web, that we could never as humans digest and read. There’s more information created in a single day than we could absorb in a lifetime, but a machine can digest it, learn from it, understand it, and dynamically build knowledge of the world that we can then leverage.” 


<strong>Shengxi Wu</strong>
Shengxi Wu

Student of NLCS Jeju
Member of NLCS Jeju Computer Science Society

You May Also Like
black farmed eyeglasses in front of laptop computer
Read More

All about insertion sort

People use “sort” in many different programs; it is a primary function when it comes to organizing a…