Tom Mitchel from Carnegie Mellon University defined machine learning in mathematically understandable terms as in 1997. He said:
“A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, as measured by P, improves with experience E.”