Can someone please provide examples of labelled and unlabelled data? I have been reading definitions of semi supervised learning but it does not make clear on what the two actually are.
Let's say you want to classify some patients in two categories: healty and sick patients. Then you can use labelled training data. Labelled data have a label, in our case:
label gender age healthy m 18 healthy f 29 healthy f 34 healthy m 21 ... sick m 68 sick f 74 sick m 65
Unlabelled data could be in our case the new patients arriving to the hospital
gender age f 65 m 21 ... m 23 f 18 f 75
Based on these labelled patients, you could try to classify the unlabelled patients. This is known as supervised learning.
But if you only have unlabelled data, you can try to define clusters of patients, i.e. groups of patients showing similarities: there are for this purpose an endless number of methods and theories. In our case, we could find that two clusters,
age>60, define our data pretty well. This is called unsupervised learning.
Now semi-supervised learning, is just that you use both labelled and unlabelled data together to categorize your patients. You can use unlabelled data to build clusters and the few labelled data points to decide which clusters represent healthy and sick patients.
Note: I use a specific categorization example for the purpose of illustrating the concept, but there are many other machine learning problems being solved through semi-supervised learning.