তাবলীগী মুরুব্বীদের নির্বাচিত বয়ান ও সহজ ছয় নম্বর (হার্ডকভার) পিডিএফ ডাউনলোড করুন

arpit_patel

The bible of learning theory, the book every machine learning theorist should have in his/her private library. Well ... let's be more precise! What is this book about? This book is about the theory of machine learning, and more specifically, the theory of regression when you do not have any strong assumption on the model of the data (so, no Gaussian assumption, no Linear assumption, etc ...). Most of the book is about proving that this specific approach is consistent and then book provides a convergence rate for that method. You can see these kinds of arguments for local averagers (K-NN, Kernel method, Partitioning), Least Squares methods (with and without regularization), and several other topics. If you are interested in theory of machine learning, you should read at least some parts of this book - even though the book is not an easy read (I plan to understand the whole book in the next few years!!!). Remember that this book is not an algorithmic book. So, you should not read it to learn or even understand "new" machine learning methods. Use Hastie, Tibshirani, and Friedman's book instead which is an excellent book on its own (though not the easiest one anyway). Also note that this book does not consider classification problems so much. However, the theory -as I partially understand it- should not be that different. If you are really interested about the theory of classification, you may like to read Deroye, Gyorfi, and Lugosi, A Probabilistic Theory of Pattern Recognition, 1997 instead.

2022-10-26 05:35