Think about some possible binary classification, multi-class
classification and regression tasks. What hypothesis spaces can be
considered?
In binary classification, hypothesis spaces can be considered a line, a circle... which separates the data into two classes. For example, classify if an email is spam or not.
The core of multi-class classification, is the same as binary classification. The difference between them is that multi-class takes into account more groups instead of only two of them. So, the hypothesis space can be several lines, curved lines or some figures. For instace, classify a child for the year of the school they belong to.
Finally, is considered in regression tasks not categorical variables but continuos variables to fix the data. The hypothesis space could be polynomial regression for any grades. For instance, it could be studied the dependence between the GDP (gross domestic product) for an economy and the microcredits given in the economy or the employment rate amount others variables.