仿生模式識別與多權(quán)值神經(jīng)元

書籍目錄

Part Ⅰ Review of Statistics Pattern Recognition Chapter 1 Introduction of Pattern Recognition 1.1 Pattern Recognition Concept 1.2 Pattern Recognition System Biasic Processes 1.3 A Brief Survey of Pattern Recognition Appro aches 1.4 Scope and Organization Chapter 2 Kernel of Statistical Pattern Recognition and Pre-Precessing 2.1 Question Arise 2. 1. 1 Question Expression 2. 1.2 Empirical Risk Minimization 2. 1.3 Generalization Ability and Complexity 2.2 Kernel of Statistical Pattern Recognition 2. 2. 1 Vapnik-Chervonenkis Dimension 2. 2. 2 The Bounds of Generalization Ability 2.2.3 The Minimization of Structural Risk 2.3 Preprocessing 2.4 Feature Extraction and Feature Selection 2.4. 1 Curse of Dimensionality  2.4. 2 Feature Extraction 2. 4.3 Feature Selection 2.5 Support Vector Manchine 2. 5.1 The Optimal Hyperplane Under Linearly Separable 2.5.2 The Soft Spacing Under Linearly Nonseparable 2. 5.3 The Kernel Function Under Non-Linear Case 2. 5.4 Support Vector Machine's Traits and Advantages References Part Ⅱ Biomimetic Pattern Recognition Chapter 3 Introduction Chapter 4 The Foundation of Biomimetic Pattern Recognition 4.1 Overview of High-Dimensional Biomimetic Informatics 4. 1.1 The Proposal of the Problem of Computer Irnaginal Thinking 4. 1.2 The Principle of High-Dimensional Biomimetic Informatics " 4.2 Basic Contents of High-Dimensional Biomimetic Informatics 4.3 Main Features of High-Dimensional Biomimetic Informatics 4.4 Concepts and Mathematical Symbols In High-Dimensional Biomimetic Informatics 4.4. 1 Concepts and Definitions 4. 4. 2 Mathematical Symbols 4. 4. 3 Symbolic Computing Methods in Resolving Geometry Computing Problems 4.4. 4 Several Applications in Solving Complicated Geometry Computing Problems 4.5 Some Applications 4. 5. 1 Blurred Image Restoration 4. 5.2 Uneven Lighting Image Correction 4. 5.3 Removing Facial Makeup Disturbances Chapter 5 The Theory of Biomimetie Pattern Recognition 5.1 The Concept of Biomimetic Pattern Recognition 5.2 The Choice of The Name 5.3 The Developments of Biomimetic Pattern Recognition 5.4 Covering. The Concept of Recognition in Biomimetic Pattern Recognition 5.5 The Principle of Homology-Continuity. The Starting Point of Biomimetic Pattern Recognition 5.6 Expansionary Product 5.7 Experiments 5.7. 1 The Architecture of the Face Recognition System 5.7.2 Umist Face Data 5.7.3 Pre-treatment 5.7. 4 The Realization of SVM Face Recognition Algorithms 5.7. 5 The Realization of BPR Face Recognition Algorithms 5.7. 6 Experiments Results and Analyzes 5.8 Summary Chapter 6 Applications 6.1 Object Recognition 6.2 A Multi-Camera Human-Face Personal Identification System 6.3 A Recognition System For Speaker-Independent Continuous Speech  6.4 Summary References Part Ⅲ Multi-Weight Neurons and Networks Chapter 7 History And Definations of Artificial Neural Networks 7.1 From Biological Neural Networks to Artificial Neural Networks and Its Development 7.2 Some Definitions and Concepts of Artificial Neural Networks 7.3 Unifications and Divergences Between Array-Processors and Neural Networks 7.4 Artificial Neural Networks' Effects on Nanoelectronical Computational Technology Chapter 8 Geometric Concepts of Artificial Neurons 8.1 Mathematical Expressions of Common Neurons and Their Geometric Concepts 8.2 General Mathematical Model of Common Neurons and Its Geometric Concept 8.3 Direction Basis Function Neuron and Its Geometric Concept 8.4 Multi-Threshold Neurons and Networks Chapter 9 Multi-Weight Neurons and Their Applications 9.1 General Mathematical Expression of Multi-Weight Neurons' Functions 9.2 Interchangeabilities of Points, Vectors, Hyper Planes in High-Dimensional Space 9.3 Effect of High-Dimensional Point Distribution Analysis in Information Technology 9.4 Multi-Weight Neurons are Computing Tools on High-Dimensional Point Distribution Analysis 9.5 Applications of Multi-Weight Neurons and Networks On Biomimetic Pattern Recognition References Appendix Experts' Evaluation to The Book

章節(jié)摘錄

版權(quán)頁：   插圖：   A child in a very small age can quickly identify their relatives, regardless of the situation whether his (her) relative is alone in the room or in the crowd;whether in sunny noon or in the dimly lit evening; and whether his or her dressing, wearing or hair-style changes. These capabilities seem so natural on children or adults. However, it is not easy to teach the computer to do similar things, despite the modern high-speed computers calculate 100 times faster than the speed of the sum of all human brains. Today's information science and artificial intelligence disciplines have developed over dozens of years, but the computers do not know how to calculate in face of imaginal thinking problems. The reason lies in the difference of imaginal thinking and logical thinking in time and space. The computer of Yon Neumann structure can be "familiar"with the logical thinking problems dealt in the left brain of human, but is "helpless" with imaginal thinking problems dealt in the right brain. Therefore,the ancient mathematical tools and calculation methods are no longer applica-ble, thus the creation and development of new disciplines is expected. 4. 1.2 The Principle of High-Dimensional Biomimetic Informatics It is well known that digital information such as an image can be thought of as a point in a high-dimensional Euclidean space in informatics. If each num-ber of a piece of data is regarded as a coordinate value, the whole group of numbers could correspond to a vector in a high-dimensional space. Usually, representing digital data as points is a kind of illustration and vi-sualization of analytic formulas, which helps people understand the classifica-tion and distribution of the sample data in high-dimensional space easily. For example, a hidden neuron in Back-Propagation (BP) Algorithm can be seen asa hyperplane, although BP Algorithm didn't come up from the geometric char-acters of hyperplanes in high-dimensional space.

編輯推薦

《仿生模式識別與多權(quán)值神經(jīng)元》由國防工業(yè)出版社出版。

圖書封面

評論、評分、閱讀與下載

---

    仿生模式識別與多權(quán)值神經(jīng)元 PDF格式下載

---