What exactly is the difference between supervised and unsupervised learning?

Question

I am trying to understand clustering methods.

What I I think I understood:

1. In supervised learning, the categories/labels data is assigned to are known before computation. So, the labels, classes or categories are being used in order to "learn" the parameters that are really significant for those clusters.

2. In unsupervised learning, datasets are assigned to segments, without the clusters being known.

Does that mean that, if I don't even know which parameters are crucial for a segmentation, I should prefer supervised learning?

Answer 1

The difference is that in supervised learning the "categories", "classes" or "labels" are known. In unsupervised learning, they are not, and the learning process attempts to find appropriate "categories". In both kinds of learning all parameters are considered to determine which are most appropriate to perform the classification.

Whether you chose supervised or unsupervised should be based on whether or not you know what the "categories" of your data are. If you know, use supervised learning. If you do not know, then use unsupervised.

As you have a large number of parameters and you do not know which ones are relevant, you could use something like principle component analysis to help determine the relevant ones.

Answer 2

Note that there are more than 2 degrees of supervision. For example, see the pages 24-25 (6-7) in the PhD thesis of Christian Biemann, Unsupervised and Knowledge-free Natural Language Processing in the Structure Discovery Paradigm, 2007.

The thesis identifies 4 degrees: supervised, semi-supervised, weakly-supervised, and unsupervised, and explains the differences, in a natural-language-processing context. Here are the relevant definitions:

• In supervised systems, the data as presented to a machine learning algorithm is fully labelled. That means: all examples are presented with a classification that the machine is meant to reproduce. For this, a classifier is learned from the data, the process of assigning labels to yet unseen instances is called classifi- cation.
• In semi-supervised systems, the machine is allowed to additionally take unlabelled data into account. Due to a larger data basis, semi-supervised systems often outperform their supervised counterparts using the same labelled examples. The reason for this improvement is that more unlabelled data enables the system to model the inherent structure of the data more accurately.
• Bootstrapping, also called self-training, is a form of learning that is designed to use even less training examples, therefore sometimes called weakly-supervised. Bootstrapping starts with a few training examples, trains a classifier, and uses thought-to-be positive examples as yielded by this classifier for retraining. As the set of training examples grows, the classifier improves, provided that not too many negative examples are misclassified as positive, which could lead to deterioration of performance.
• Unsupervised systems are not provided any training examples at all and conduct clustering. This is the division of data instances into several groups. The results of clustering algorithms are data driven, hence more ’natural’ and better suited to the underlying structure of the data. This advantage is also its major drawback: without a possibility to tell the machine what to do (like in classification), it is difficult to judge the quality of clustering results in a conclusive way. But the absence of training example preparation makes the unsupervised paradigm very appealing.

Answer 3

In supervised learning the classes are known in advance and also their types, for instance, two classes good and bad customers. When new object(customer) comes on the basis of its attributes the customer can be assigned to bad or good customer class.

In unsupervised learning the groups/classes are not already known, we have objects (customers), so group the customers having similar buying habits hence different groups are made of the customers i.e. not known already on the basis of similar habits of buying.

Answer 4

In supervised learning the output(dependent variable) depends on the input variable(independent variable).In some set of given supervisions the responder tries to compute the desired objective.

In unsupervised learning there is no supervision so system tries to adapt itself to the situation and learns manually based on some measure.

eg: Teacher in a class -supervision -supervised learning An self study elective in class-No supervision Unsupervised Learning