Advice for Students of Machine Learning written by David Mimno, assistant professor of Information Science at Cornell University, Mr. Mimno summarizes some suggestions for students/beginners of machine learning, emphasizing the combination of theory and practice, patience and perseverance. The method of in-depth study of a paper is also proposed. Hope to be helpful to readers. The original address: www.mimno.org/articles/ml… Mr Mimno homepage: mimno.infosci.cornell.edu/ below is the translation text:

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = title: advice to machine learning for beginners Author: David Mimno translator: whatbeg

Recently one of my students asked me for advice on learning machine learning, so I wrote this article, which is biased towards my own experience, but I’m going to generalize anyway.

My favorite book to start with right now is Kevin Murphy’s book Machine Learning, which you might want to read

  • Pattern Recognition and Machine Learning by Chris Bishop
  • Probabilistic Graphical Models: Principles and Techniques by Daphne Koller
  • And Information Theory, Inference and Learning Algorithms by David Mackay.

Anything you learn about linear algebra and probability theory/statistics will help.

  • Introduction To Linear Algebra by Strang
  • Bayesian Data Analysis, Second Edition by Gelman,Carlin,Stern, and Rubin
  • Gelman and Hill’s Data Analysis Using Regression and Multilevel/Hierarchical Models are some of my favorite books.

Instead of trying to get all aspects of a particular knowledge at the beginning, try to read descriptions of the same knowledge from many different sources.

Nothing is more important than trying for yourself. Pick a model, implement it, compare it to other open source implementations, and think, are there any computational or mathematical tricks that make the program work?

Read some papers. When I was in graduate school, I spent 20 minutes on the bus in the morning and 20 minutes in the evening, and I always had an interesting paper in my bag. The bus wasn’t the point — what really mattered was that I was able to spend about half an hour a day reading [the papers].

Pick a paper you like, and spend a week digging into it, thinking about it all the time, remembering the form of each formula, taking a walk and thinking about how each variable affects the outcome, how the different variables affect each other. Think about how Formula 6 leads to formula 7 — the authors often omit intermediate steps and algebraic details, so you can complete the steps.

Be patient and persistent. Remember von Neumann, “Young man, in mathematics you don’t understand things. You just get used to them.” (Basically, young man, you don’t understand anything in math, you just get used to it.)