How do I learn machine learning?

• Start from Series of videos ::https://www.youtube.com/view_play_list?p=A89DCFA6ADACE599

•  There were plenty of ways to learn Machine Learning / data mining.
  1) Primary thing is to select and learn most suitable language for data mining. ex: python, R.
  I prefer python.
  
  2)Learning some packages like numpy,pandas,matplotlib,scikit-learn.
  
  3) Just follow any reference book for data mining / Machine Learning.
  
  4) Just learning the algorithm will never make you perfect person in data mining or Machine learning field you have to apply them for some data sets. you can get data sets from internet itself.
  
  5) If you face any problem while applying your algorithm to data sets. Just go to A Programmer's Guide to Data Mining here you can find some very good example of how to apply data mining algorithms to data sets and with some real world  applications. I promise you if your an  really  datamining / Machine learning  aspirant you will enjoy A Programmer's Guide to Data Mining.
  
  6) Once you completed this now you can search for data sets and you can apply your algorithms.
  
  7) You can spent some time on  Coursera. where you will find some coursers belongs to data mining and Machine learning. (you will get videos and problem to practice)  
  
  8) On a final note i can say that quora + google will give you lots of stuff related to data mining.
  
  9) All the best ,have a blast and have fun play with data mining / Machine Learning .
  
  10) You can check this blog once a week.
  dataaspirant

• See this talk by Jeremy Howard : At Kaggle, It’s a Disadvantage To Know Too Much and “Getting In Shape For The Sport Of Data Science”
• Start by practicing on toy datasets in MATLAB and walking through simple examples in Statistics and Neural Network Toolbox
• Below two books are standard introduction texts. They are complementary,   the first one is written from a statistician perspective with lots of data analysis examples and the second one is focusing on algorithms. Both are graduate level texts requiring knowledge of Linear algebra, Probability and statistics and Multivariable calculus. It would also help to take a  class on Optimization but not strictly necessary. The texts are:
• 1)  Hastie, Tibshirani & Friedman, The elements of statistical learning : http://www-stat.stanford.edu/~ti... 
• 2) Bishop, Pattern Recognition and Machine Learning: http://research.microsoft.com/en...

• For a more gentle read you may want to check out the textbook by T.Mitchell
• The most coherent and concise lecture notes are by Andrew Ng,   also see his ML course

• For Natural Language applications see Michael Collins'  NLP course and notes
• For Graphical Models see Daphne Koller's course and book
• For Deep Learning see NN + UFLDL + CNN courses