神经网络与机器学习 PDF 分享
神经网络是计算智能和机器学习的一个重要分支，在许多领域都取得了巨大的成功。在众多神经网络著作中，西蒙·海金的《神经网络原理》（第三版改名为《神经网络与机器学习》）影响广泛。在本书中，作者基于近年来神经网络和机器学习的新进展，从理论和实际应用出发，全面系统地介绍了神经网络的基本模型、方法和技术，并将神经网络和机器学习有机地结合起来。
“神经网络与机器学习”不仅关注数学分析方法和理论的讨论，而且非常关注神经网络在实际工程问题中的应用，如模式识别、信号处理和控制系统。这本书可读性很强。作者对神经网络的基本模型和主要学习理论进行了深入的讨论和分析，通过大量的实验报告、实例和练习，帮助读者更好地学习神经网络。
本版在前一版的基础上进行了广泛修订，提供了对两个日益重要的学科，神经网络和机器学习的新分析。
本书的特点：
1、基于随机梯度下降的在线学习算法；小规模和大规模学习问题。
2、核方法，包括支持向量机和表达式定理。
3、信息论学习模型，包括连接、独立分量分析、一致独立分量分析和信息瓶颈。
4、随机动态规划，包括近似和神经动态规划。
5、连续状态估计算法，包括卡尔曼滤波和粒子滤波。
6、使用连续状态估计算法训练递归神经网络。
7、有洞察力的计算机导向实验。
Neural network is an important branch of computational intelligence and machine learning, and has achieved great success in many fields. Among many neural network works, Simon Haykin’s principles of neural networks (the third edition is renamed neural networks and machine learning) has a wide influence. In this book, the author comprehensively and systematically introduces the basic models, methods and technologies of neural networks based on the new progress of neural networks and machine learning in recent years, starting from theory and practical application, and organically combines neural networks and machine learning.
“Neural network and machine learning” not only pays attention to the discussion of mathematical analysis methods and theories, but also pays great attention to the application of neural network in practical engineering problems such as pattern recognition, signal processing and control system. This book is very readable. The author thoroughly discusses and analyzes the basic model and main learning theories of neural networks, and helps readers learn neural networks better through a large number of experimental reports, examples and exercises.
This edition has been extensively revised on the basis of the previous edition, providing new analysis of two increasingly important disciplines, neural networks and machine learning.
Features of this book:
- Online learning algorithm based on random gradient descent; Small scale and large-scale learning problems.
- Kernel method, including support vector machine and expression theorem.
- Information theory learning model, including connection, independent component analysis (ICA), consistent independent component analysis and information bottleneck.
- Stochastic dynamic programming, including approximation and neural dynamic programming.
- Successive state estimation algorithm, including Kalman filter and particle filter.
- Use the successive state estimation algorithm to train the recurrent neural network.
- Insightful computer oriented experiments.
神经网络与机器学习 PDF 分享
原创文章受到原创版权保护。转载请注明出处：https://www.johngo689.com/5176/
转载文章受原作者版权保护。转载请注明原作者出处！
