classification -> dividing the data in categories where output classes are known. -> supervised task
regression -> output data is continous -> supervised task.
clustering -> divide the data in some groups -> unsupervised task
Anomaly detection -> find out some out of the place data point from input data -> output classes labels are not required -> unsupervised task.
Dimension reduction -> map a higher dimension input to a lower dimension input -> output classes labels are not required -> unsupervised task