000			02126nam a2200217Ia 4500
003			OSt
005			20260427155722.0
008			210219s9999||||xx |||||||||||||| ||und||
020			_a9780262043793
082			_a006.31 ALP-I
100			_aAlpaydin, Ethem
245		0	_aIntroduction to machine learning
250			_a4th
260			_aCambridge, Massachusetts _bMIT Press _c2020
300			_axxiv, 682 pages : illustrations ; 24 cm.
520			_aSince the third edition of this text appeared in 2014, most recent advances in machine learning, both in theory and application, are related to neural networks and deep learning. In this new edition, the author has extended the discussion of multilayer perceptrons. He has also added a new chapter on deep learning including training deep neural networks, regularizing them so they learn better, structuring them to improve learning, e.g., through convolutional layers, and their recurrent extensions with short-term memory necessary for learning sequences. There is a new section on generative adversarial networks that have found an impressive array of applications in recent years. Alpaydin has also extended the chapter on reinforcement learning to discuss the use of deep networks in reinforcement learning. There is a new section on the policy gradient method that has been used frequently in recent years with neural networks, and two additional sections on two examples of deep reinforcement learning, which both made headlines when they were announced in 2015 and 2016 respectively. One is a network that learns to play arcade video games, and the other one learns to play Go. There are also revisions in other chapters reflecting new approaches, such as embedding methods for dimensionality reduction, and multi-label classification. In response to requests from instructors, this new edition contains two new appendices on linear algebra and optimization, to remind the reader of the basics of those topics that find use in machine learning
650			_aMachine learning _976662
650			_aArtificial Intelligence _976663
942			_cBK
999			_c199921 _d199921