A good indespensible textbook for me had been chris bishops book Pattern Recognition and Machine Learning. However it might be suited best for the grad student.

A good indespensible textbook for me had been chris bishops book Pattern Recognition and Machine Learning. However it might be suited best for the grad student. Honestly the best thing you can do is start working on machine learning problems. Try and build a classifier or predictor. Brush up on probability theory. If you have no idea where to begin, check out Collective Intelligence. It's a great intro read that gives a good background into the types of problems ML and AI are trying to solve.