دانلود کتاب مبانی یادگیری عمیق

This book is concerned with the concepts, theories, and algorithms of deep learning. In Part I, we provide the fundamental knowledge about deep learning by exploring the traditional supervised and unsupervised learning algorithms and the alternative advanced machine learning type, called ensemble learning. In Parts II and III,we describe the modification of existing machine learning algorithms into deep versions and the deep neural networks, such as Multiple Layer Perceptron, Recurrent Neural Networks, Restricted Boltzmann Machine, and Convolutional Neural Networks, respectively. In Part IV, we cover the specialization of deep learning algorithms into the text mining tasks. Readers need the basic level of knowledge about linear algebra, vector calculus, and traditional machine learning algorithms for understanding the deep learning algorithms.
We mention some agenda which provide motivations for writing this book. The machine learning algorithms became very popular approaches to data mining tasks, because of their flexibility and adaptability, compared with the rule-based approaches. The deep learning algorithms are derived from the machine learning algorithms and become the advanced approaches to nonlinear problems. The deep learning algorithms are applied to tasks in various disciplines such as control system engineering, chemical engineering, and business administration and management. In this book, we describe the deep learning algorithms with the assumption that the input data is given as a numerical vector.
This book consists of the four parts: foundation, deep machine learning, deep neural networks, and textual deep learning. In the first part, we cover the swallow supervised learning, the swallow unsupervised learning, and the ensemble learning as background for understanding the deep learning. In the second part, we modify the four kinds of supervised machine learning algorithms, K-Nearest Neighbor, Probabilistic Learning, the Decision Tree, and the SVM (Support Vector Machine), into their deep learning versions. In the third part, we describe some typical deep learning algorithms: MLP (Multiple Layer Perceptron), Recurrent Neural Networks, Restricted Boltzmann Machine, and Convolutional Neural Networks. In the last part, we specialize some deep operations on numerical values into ones on textual data; the text summarization is used to extract some of each article, like the pooling.