دانلود کتاب تحلیل و طراحی شبکه‌های عصبی تأخیری

Artificial neural network (NN) is often called NN for short. The dynamics of NNs have been applied to multiple fields, such as associative memory, autonomous navi-gation, data encryption, intelligent control, medicine diagnosis, optimization, pattern recognition, and so forth. Roughly speaking, mathematical models of NNs are divided into the continuous-time ones and the discrete-time ones, which are described by differential equations and difference equations, respectively. Delays resulting from the communication mode among neurons are classified as transmission delay, leakage delay, and distributed delay. Since delay is inevitable, it is necessary to model NNs by functional differential/difference equations, that is, differential/difference equations with delayed states. As a result, many experts and scholars have paid attention to the study of delayed NNs, and many excellent results have been reported in literature. From the point of control theory, the research on delayed NNs includes mainly two aspects: analysis and design. The current studies mainly focus on the continuous-time case. However, digital computers process only discrete data, so discrete-time NNs can match compute performance better from the view of digitization than contin-uous ones. Thus, it is equally important to research the discrete-time NNs and the continuous-time ones.
It was apparent to all that the so-called Lyapunov–Krasovskii functional (LKF) method is most commonly employed to research the problems of analysis and design of NNs with delays. The merits of LKF method lie mainly in: (i) it can be applied to almost all problems on analysis and design of NNs with delays; and (ii) it is easily mastered and applied by a beginner. However, in general, no united technique to give an effective and appropriate LKF. Moreover, sufficient conditions derived from the LKF method contain generally multiple linear matrix inequalities (LMIs) and decision variables, and they are increasing as delays grow in number, which will result in a huge growth in the amount of computation. Therefore, it is significative to give a novel method for analysis and design of delayed NNs.
The aim of this book is to present a direct method based on system solutions to address the problems of analysis and design of NNs with delays, which has the following advantages:
1. It does not involve the construction of any LKFs, which overcomes the difficulty in finding a suitable LKF;
2. It can provide simpler sufficient conditions, and hence it possesses less compu-tational complexity;
3. It can result in delay-dependent global exponential stability criteria that can offer the decay rate estimation of the state;
4. It is suitable for analysis and design problems of most system models with (many kinds of) delays after a small modification.

The whole book includes 11 chapters and involves the problems of analysis and design of NNs with delays. All the contents are taken from the authors’ publications during the past 5 years.
The book is written for graduate students and mathematicians at the research level. It can also become a reference for postgraduate students.
Many experts and scholars in the world have made outstanding contributions to the analysis and design of NNs with delays. Due to the limitation of length, we cannot collect all these contributions in the book. I would like to apologize to these experts and scholars.