دانلود کتاب شبکه‌های عصبی کاربردی در پایتون و متلب

The inception of neural network research dates to the 1940s with the seminal collab-oration between psychologist Warren Sturgis McCulloch and mathematician Walter Pitts. They introduced the McCulloch-Pitts (MP) model, which laid the groundwork for subsequent neural network research. The evolution of neural networks can be charted through three distinct phases: the initial phase from 1947 to 1969, which was a period of prolific model and rule proposals, including the MP model, Hebb’s learning rules, and the perceptron model; the transitional phase from 1970 to 1986, following a lull in research activity, which saw significant contributions such as John Hopfield’s introduction of energy functions to networks, providing stability criteria and pathways for associative memory and optimal computation. In 1984, Geoffrey Hinton proposed the Boltzmann machine model, and in 1986, David Rumelhart, Geoffrey Hinton, and Ronald Williams introduced the error backpropagation (BP) neural network, which has since become a cornerstone in the field. The current phase, from 1987 to the present, marks a period of global attention and research, propelling neural networks to new heights of development. Neural networks are characterized by several distinctive features: 1.High fault tolerance due to the distributed storage of information across network neurons. 2.Rapid calculations facilitated by parallel processing methods. 3.Self-learning, self-organization, and self-adaptability enable the network to handle uncertain or unknown systems. 4.The ability to approximate any complex non-linear relationships, with strong information synthesis capabilities, managing both quantitative and qualitative information, coordinating multiple input information relationships, and being well-suited for multi-information fusion and multimedia technology.