دسترسی نامحدود
برای کاربرانی که ثبت نام کرده اند
برای ارتباط با ما می توانید از طریق شماره موبایل زیر از طریق تماس و پیامک با ما در ارتباط باشید
در صورت عدم پاسخ گویی از طریق پیامک با پشتیبان در ارتباط باشید
برای کاربرانی که ثبت نام کرده اند
درصورت عدم همخوانی توضیحات با کتاب
از ساعت 7 صبح تا 10 شب
ویرایش: 1 نویسندگان: Licheng Jiao, Ronghua Shang, Fang Liu, Weitong Zhang سری: ISBN (شابک) : 0128197951, 9780128197950 ناشر: Elsevier Science Ltd سال نشر: 2020 تعداد صفحات: 763 زبان: English فرمت فایل : PDF (درصورت درخواست کاربر به PDF، EPUB یا AZW3 تبدیل می شود) حجم فایل: 74 مگابایت
در صورت تبدیل فایل کتاب Brain and Nature-inspired Learning, Computation and Recognition به فرمت های PDF، EPUB، AZW3، MOBI و یا DJVU می توانید به پشتیبان اطلاع دهید تا فایل مورد نظر را تبدیل نمایند.
توجه داشته باشید کتاب مغز و یادگیری ، محاسبه و شناخت با الهام از طبیعت نسخه زبان اصلی می باشد و کتاب ترجمه شده به فارسی نمی باشد. وبسایت اینترنشنال لایبرری ارائه دهنده کتاب های زبان اصلی می باشد و هیچ گونه کتاب ترجمه شده یا نوشته شده به فارسی را ارائه نمی دهد.
یادگیری، محاسبات و شناخت الهام گرفته از مغز و طبیعت تجزیه و تحلیل سیستماتیک شبکه های عصبی، محاسبات طبیعی، یادگیری ماشین و فشرده سازی، الگوریتم ها و برنامه های کاربردی الهام گرفته از مغز و مکانیسم های بیولوژیکی موجود در طبیعت را ارائه می دهد. بخش ها پیشرفت های جدید و برنامه های کاربردی اصلی، الگوریتم ها و شبیه سازی ها را پوشش می دهند. تحولات در یادگیری الهام گرفته از مغز و طبیعت علاقه به پردازش تصویر، مشکلات خوشهبندی، تشخیص تغییر، نظریه کنترل و سایر رشتهها را افزایش داده است. این کتاب مشکلات و کاربردهای اصلی مربوط به محاسبات و تشخیص الهام گرفته از زیستی، معرفی پیادهسازی الگوریتم، شبیهسازی مدل و کاربرد عملی تنظیم پارامتر را مورد بحث قرار میدهد.
خوانندگان راه حل هایی برای مشکلات در محاسبات و تشخیص، به ویژه شبکه های عصبی، محاسبات طبیعی، یادگیری ماشین و سنجش فشرده پیدا خواهند کرد. این جلد مقدمه ای جامع و ساختار یافته برای یادگیری، محاسبات و شناخت الهام گرفته از مغز و طبیعت ارائه می دهد.
Brain and Nature-Inspired Learning, Computation and Recognition presents a systematic analysis of neural networks, natural computing, machine learning and compression, algorithms and applications inspired by the brain and biological mechanisms found in nature. Sections cover new developments and main applications, algorithms and simulations. Developments in brain and nature-inspired learning have promoted interest in image processing, clustering problems, change detection, control theory and other disciplines. The book discusses the main problems and applications pertaining to bio-inspired computation and recognition, introducing algorithm implementation, model simulation, and practical application of parameter setting.
Readers will find solutions to problems in computation and recognition, particularly neural networks, natural computing, machine learning and compressed sensing. This volume offers a comprehensive and well-structured introduction to brain and nature-inspired learning, computation, and recognition.
Brain and Nature-Inspired Learning, Computation and Recognition Copyright 1. Introduction 1.1 A brief introduction to the neural network 1.1.1 The development of neural networks 1.1.2 Neuron and feedforward neural network 1.1.3 Backpropagation algorithm 1.1.4 The learning paradigm of neural networks 1.2 Natural inspired computation 1.2.1 Fundamentals of nature-inspired computation 1.2.2 Evolutionary algorithm 1.2.3 Artificial immune system (AIS) 1.2.4 Other methods 1.3 Machine learning 1.3.1 Development of machine learning 1.3.2 Dimensionality reduction 1.3.3 Sparseness and low-rank 1.3.4 Semisupervised learning 1.4 Compressive sensing learning 1.4.1 The development of compressive sensing 1.4.2 Sparse representation 1.4.3 Compressive observation 1.4.4 Sparse reconstruction 1.5 Applications 1.5.1 Community detection 1.5.2 Capacitated arc routing optimization 1.5.3 Synthetic aperture radar image processing 1.5.4 Hyperspectral image processing References 2. The models and structure of neural networks 2.1 Ridgelet neural network 2.2 Contourlet neural network 2.2.1 Nonsubsampled contourlet transforms 2.2.2 Deep contourlet neural network 2.3 Convolutional neural network 2.3.1 Convolution 2.3.2 Pooling 2.3.3 Activation function 2.3.4 Batch normalization 2.3.5 LeNet5 2.4 Recurrent artificial neural network 2.5 Generative adversarial nets 2.5.1 Biological description—human behavior 2.5.2 Data augmentation 2.5.3 Model description 2.6 Autoencoder 2.6.1 Layer-wise pretraining 2.6.2 Autoencoder network 2.7 Restricted Boltzmann machine and deep belief network Further reading 3. Theoretical basis of natural computation 3.1 Evolutionary algorithms 3.1.1 Pattern theorem 3.1.2 Implicit parallelism 3.1.3 Building block assumption 3.2 Artificial immune system 3.2.1 Markov chain-based convergence analysis 3.2.2 Nonlinear dynamic model 3.3 Multiobjective optimization 3.3.1 Introduction 3.3.2 Mathematical concepts 3.3.3 Multiobjective optimization algorithms 3.3.3.1 The first generation of evolutionary multiobjective optimization algorithms 3.3.3.1.1 MOGA 3.3.3.1.2 NSGA 3.3.3.1.3 NPGA 3.3.3.2 The second generation of evolutionary multiobjective optimization algorithms 3.3.3.2.1 SPEA and SPEA2 3.3.3.2.2 PAES, PESA, and PESA-II 3.3.3.2.3 NSGA-II References 4. Theoretical basis of machine learning 4.1 Dimensionality reduction 4.1.1 Subspace segmentation 4.1.2 Nonlinear dimensionality reduction 4.2 Sparseness and low rank 4.2.1 Sparse representation 4.2.2 Matrix recovery and completion 4.3 Semisupervised learning and kernel learning 4.3.1 Semisupervised learning 4.3.2 Nonparametric kernel learning References 5. Theoretical basis of compressive sensing 5.1 Sparse representation 5.1.1 Stationary dictionary 5.1.2 Learning dictionary 5.2 Compressed observation 5.3 Sparse reconstruction 5.3.1 Relaxation methods 5.3.2 Greedy methods 5.3.3 Natural computation methods 5.3.4 Other methods References 6. Multiobjective evolutionary algorithm (MOEA)-based sparse clustering 6.1 Introduction 6.1.1 The introduction of MOEA on constrained multiobjective optimization problems 6.1.2 An introduction to MOEA on clustering learning and classification learning 6.1.3 The introduction of MOEA on sparse spectral clustering 6.2 Modified function and feasible-guiding strategy-based constrained MOPs 6.2.1 Problem description 6.2.2 Modified objective function 6.2.3 The feasible-guiding strategy 6.2.4 Procedure for the proposed algorithm 6.3 Learning simultaneous adaptive clustering and classification learning via MOEA 6.3.1 Objective functions of MOASCC 6.3.2 The framework of MOASCC 6.3.3 Computational complexity 6.4 A sparse spectral clustering framework via MOEA 6.4.1 Mathematical description of SRMOSC 6.4.2 Extension on semisupervised clustering 6.4.3 Initialization 6.4.4 Crossover 6.4.5 Mutation 6.4.6 Laplacian matrix construction 6.4.7 Final solution selection phase 6.4.8 Complexity analysis 6.5 Experiments 6.5.1 The experiments of MOEA on constrained multiobjective optimization problems 6.5.1.1 Experimental setup 6.5.1.2 Performance metrics 6.5.1.2.1 IGD 6.5.1.2.2 Minimal spacing 6.5.1.2.3 Coverage of two sets (ς) 6.5.1.3 Comparison experiment results 6.5.2 The experiments of MOEA on clustering learning and classification learning 6.5.2.1 Experiment setup 6.5.2.2 Experiment on a synthetic datasets 6.5.2.3 Experiment on real-life datasets 6.5.3 The experiments of MOEA on sparse spectral clustering 6.5.3.1 Detailed analysis of SRMOSC 6.5.3.2 Experimental comparison between SRMOSC and other algorithms 6.6 Summary References 7. MOEA-based community detection 7.1 Introduction 7.2 Multiobjective community detection based on affinity propagation 7.2.1 Background to APMOEA 7.2.1.1 Affinity propagation method 7.2.1.2 Multiobjective optimization 7.2.2 Objective functions 7.2.3 The selection method for nondominated solutions 7.2.4 Preliminary partition by the AP method 7.2.5 Further search using multiobjective evolutionary algorithm 7.2.5.1 Representation and initialization 7.2.5.2 Genetic operators 7.2.6 Elitist strategy of the external archive 7.3 Multiobjective community detection based on similarity matrix 7.3.1 Background of GMOEA-net 7.3.1.1 Structural balance theory 7.3.1.2 Tchebycheff approach 7.3.2 Objective functions 7.3.3 The construction of similarity matrix and k-nodes update policy 7.3.3.1 The function of node similarity 7.3.3.2 The k-nodes update policy 7.3.4 Evolutionary operators 7.3.4.1 The cross-merging operator based on local node sets 7.3.4.2 The mutation operator based on similarity matrix 7.3.5 The whole framework of GMOEA-net 7.4 Experiments 7.4.1 Evaluation index 7.4.2 Networks for simulation 7.4.2.1 Computer-generated networks 7.4.2.2 Real-world networks 7.4.3 Comparison algorithms and parameter settings 7.4.3.1 Comparison algorithms 7.4.3.2 Parameter settings 7.4.4 Experiments on computer-generated networks 7.4.4.1 Experiments on APMOEA 7.4.4.2 Experiments on GMOEA-net 7.4.5 Experiments on real-world networks 7.5 Summary References 8. Evolutionary computation-based multiobjective capacitated arc routing optimizations 8.1 Introduction 8.2 Multipopulation cooperative coevolutionary algorithm 8.2.1 Related works 8.2.1.1 The model of MO-CARP 8.2.1.2 The description of direction vector 8.2.2 Initial population and subpopulations partition 8.2.3 The fitness evaluation in each subpopulation 8.2.4 The elitism archiving mechanism 8.2.4.1 The external elitism archive 8.2.4.2 The internal elitism archive 8.2.5 The cooperative coevolutionary process 8.2.5.1 Construct evolutionary pool for each subregion 8.2.5.2 Crossover 8.2.5.3 Local search 8.2.5.4 The selection of offspring solutions and diversity preservation mechanism 8.2.6 The processing flow of MPCCA 8.3 Immune clonal algorithm via directed evolution 8.3.1 Antibody initialization 8.3.2 Immune clonal operation 8.3.3 Immune gene operations 8.3.3.1 The decomposition operation of the population 8.3.3.2 Gene recombination operator 8.3.3.3 Gene mutation operator 8.3.3.4 Directed comparison operator 8.3.3.5 Clonal selection operator 8.3.4 The processing flow of DE-ICA 8.4 Improved memetic algorithm via route distance grouping 8.4.1 Solutions for the timely replacement of IRDG-MAENS 8.4.2 Determine the regions which individuals belong to 8.4.3 The processing flow of IRDG-MAENS 8.5 Experiments 8.5.1 Test problems and experimental setup 8.5.1.1 MPCCA 8.5.1.2 DE-ICA 8.5.1.3 IRDG-MAENS 8.5.2 The performance metrics 8.5.2.1 The distance to the reference set (ID) 8.5.2.2 Purity 8.5.2.3 Hypervolume (HV) 8.5.3 Wilcoxon signed rank test 8.5.4 Comparison of the evaluation metrics 8.5.4.1 MPCCA 8.5.4.2 DE-ICA 8.5.4.3 IRDG-MAENS 8.5.5 Comparison of nondominant solutions 8.5.5.1 MPCCA 8.5.5.2 DE-ICA 8.5.5.3 IRDG-IDMAENS 8.6 Summary References 9. Multiobjective optimization algorithm-based image segmentation 9.1 Introduction 9.2 Multiobjective evolutionary fuzzy clustering with MOEA/D 9.2.1 Fuzzy-C means clustering algorithms with local information 9.2.2 Framework of MOEFC 9.2.3 Opposition-based learning operator 9.2.4 Mixed population initialization 9.2.5 The time complexity analysis 9.3 Multiobjective immune algorithm for SAR image segmentation 9.3.1 Definitions of AIS-based, multiobjective optimization 9.3.2 The stage of features extraction and preprocessing 9.3.2.1 Watershed raw segmentation 9.3.2.2 Feature extraction using Gabor filters and GLCP 9.3.3 The immune multiobjective framework for SAR imagery segmentation 9.4 Experiments 9.4.1 The MOEFC experiments 9.4.1.1 Experimental setting of MOEFC 9.4.1.2 Segmentation results on synthetic images 9.4.1.3 Segmentation results on natural images 9.4.1.4 Segmentation results on medical images 9.4.1.5 Segmentation results on SAR images 9.4.2 The IMIS experiments 9.4.2.1 IMIS experimental settings 9.4.2.2 Analysis of experimental results 9.5 Summary References 10. Graph-regularized feature selection based on spectral learning and subspace learning 10.1 Nonnegative spectral learning and subspace learning-based graph-regularized feature selection 10.1.1 Dual-graph nonnegative spectral learning 10.1.2 Dual-graph sparse regression 10.1.3 Feature selection 10.1.4 Optimization 10.1.5 Local structure preserving 10.1.6 Update rules for SGFS 10.2 Experiments of spectral learning and subspace learning methods for feature selection 10.2.1 Experiments and analysis of NSSRD 10.2.1.1 Experimental settings 10.2.1.2 Simple illustrative example problem 10.2.1.3 Evaluating the effectiveness of NSSRD 10.2.1.4 Clustering results and analysis 10.2.2 Experiments and analysis of SGFS 10.2.2.1 Experimental setting 10.2.2.2 Convergence test 10.2.2.3 AT&T face dataset example 10.2.2.4 Experimental results and analysis 10.2.2.5 Robustness test of algorithms 10.2.2.6 Parameter sensitivity analysis References 11. Semisupervised learning based on nuclear norm regularization 11.1 Framework of semisupervised learning (SSL) with nuclear norm regularization 11.1.1 A general framework 11.1.2 Nuclear norm regularized model 11.1.3 Modified fixed point algorithm 11.1.4 Implementation 11.1.5 Label propagation 11.1.6 Valid kernel 11.2 Experiments and analysis 11.2.1 Compared algorithms and parameter settings 11.2.2 Synthetic data 11.2.3 Real-world data sets 11.2.4 Transduction classification results References 12. Fast clustering methods based on learning spectral embedding 12.1 Learning spectral embedding for semisupervised clustering 12.1.1 Graph construction and spectral embedding 12.1.1.1 Symmetry-favored graph 12.1.1.2 Spectral embedding of graph Laplacian 12.1.2 Problem formulation 12.1.2.1 The unit hypersphere 12.1.2.2 Squared loss model 12.1.2.3 Hinge loss model 12.1.2.4 Clustering 12.1.3 Algorithm 12.1.4 Experiments 12.1.4.1 Parameter selection 12.1.4.2 Vector-based clustering 12.1.4.3 Graph-based clustering 12.2 Fast semisupervised clustering with enhanced spectral embedding 12.2.1 Problem formulation 12.2.1.1 Objective function 12.2.1.2 Solving the objective function 12.2.1.3 Clustering 12.2.2 Algorithm 12.2.2.1 Experimental results 12.2.2.2 Parameter selection 12.2.2.3 Toy examples 12.2.2.4 Vector-based clustering 12.2.2.5 Graph-based clustering References Chapter 13 - Fast clustering methods based on affinity propagation and density weighting 13.1 The framework of fast clustering methods based on affinity propagation and density weighting 13.1.1 Related works 13.1.1.1 AP clustering 13.1.1.2 Spectral clustering 13.1.1.3 Nyström method 13.1.1.4 Local length and global distance 13.1.2 Fast AP algorithm 13.1.2.1 Coarsening phase 13.1.2.1.1 Fast sampling algorithm 13.1.2.1.2 Determine the number of representative exemplars 13.1.2.2 Exemplar-clustering phase 13.1.2.3 Refinement phase 13.1.3 Fast two-stage spectral clustering framework 13.1.3.1 Fast two-stage AP algorithm 13.1.3.2 Determine the number of representative exemplars 13.1.3.3 Sampling phase 13.1.3.4 Fast-weighted approximation spectral clustering phase 13.1.3.5 Robustness 13.1.3.6 Fast nearest-neighbors research 13.2 Experiments and analysis 13.2.1 Experiments on the method based on affinity propagation 13.2.1.1 Synthetic data sets 13.2.1.2 Compared algorithms and parameter settings 13.2.1.3 Vector-based clustering 13.2.1.4 Evaluation metrics 13.2.1.5 Experimental results 13.2.1.6 Graph-based clustering 13.2.2 Experiments on the method based on density-weighting 13.2.2.1 Intertwined spirals data set 13.2.2.2 Real-world data sets 13.2.2.3 Compared algorithms 13.2.2.4 Algorithm performances 13.2.2.5 Spectral embedding References 14. SAR image processing based on similarity measures and discriminant feature learning 14.1 SAR image retrieval based on similarity measures 14.1.1 Semantic classification and region-based similarity measures 14.1.1.1 Semisupervised learning 14.1.1.2 Classification recovery scheme 14.1.1.3 Improved integrated region matching measure 14.1.1.3.1 Self-adapting k-means segmentation 14.1.1.3.2 Region-based IRM distance computation 14.1.1.3.3 Improved IRM scheme 14.1.1.3.4 Edge regions calculation 14.1.1.3.5 IIRM computation 14.1.1.4 Methodology summary 14.1.1.4.1 Off-line process 14.1.1.4.2 On-line process 14.1.1.5 Experiment 14.1.1.5.1 Performance of improved integrated region matching (IIRM) measure 14.1.1.5.2 Query example (proposed method, IRM, one of the latest retrieval methods) 14.1.1.5.3 Land cover statistical analysis 14.1.2 Fusion similarity-based reranking for SAR image retrieval 14.1.2.1 Fusion similarity-based reranking 14.1.2.1.1 Preprocessing 14.1.2.1.2 Reranking 14.1.2.1.2.1 Modal-image matrix construction and fusion similarity calculation 14.1.2.1.2.2 Reranking function and solution 14.1.2.2 Experiments and discussion 14.1.2.2.1 Experiment settings 14.1.2.2.2 Numerical assessment 14.1.2.2.2.1 Based on different retrieval methods 14.1.2.2.2.2 Compared with different reranking algorithms 14.1.2.3 Influence of different parameters 14.1.2.4 Reranking efficiency 14.1.2.5 Reranking examples 14.1.3 SAR image content retrieval based on fuzzy similarity and relevance feedback 14.1.3.1 Region-based fuzzy matching 14.1.3.1.1 Introduction to the improved integrated region matching algorithm 14.1.3.1.2 RFM measure 14.1.3.1.2.1 Superpixel-based segmentation for brightness-texture regions 14.1.3.1.2.2 Multiscale edge detector-based segmentation for edge regions 14.1.3.1.2.3 Fuzzy region representation 14.1.3.1.2.4 RFM similarity calculation 14.1.3.1.2.5 RFM summarization and computational complexity 14.1.3.2 Multiple relevance feedback (MRF) 14.1.3.3 Experiments and discussion 14.1.3.3.1 Setting parameters 14.1.3.3.2 Evaluation criteria 14.1.3.3.3 Retrieval examples 14.1.3.4 Numerical evaluation 14.1.3.4.1 Performance of the RFM 14.1.3.4.2 Performance of the proposed retrieval method 14.1.3.4.3 Importance of the multiple RF schemes\' integration 14.1.3.4.4 Significance of the RFM Gaussian kernel 14.1.3.4.5 Influences of different parameters 14.2 SAR image change detection based on spatial coding and similarity 14.2.1 Saliency-guided change detection for SAR imagery using a semisupervised Laplacian SVM 14.2.1.1 Learning a pseudotraining set via saliency detection 14.2.1.2 Obtaining change result via Laplacian support vector machine 14.2.1.3 Experimental results 14.2.1.3.1 Description of data sets 14.2.1.3.2 Quantitative analysis 14.2.1.3.3 Parameter selection 14.2.1.3.4 Experiment results and analysis on three data sets 14.2.2 SAR images change detection based on spatial coding and nonlocal similarity pooling 14.2.2.1 Producing the difference image 14.2.2.2 Learning dictionary via affinity propagation 14.2.2.3 Creating feature vectors via sparse coding and nonlocal similarity pooling 14.2.2.3.1 Obtaining a change map by k-means clustering 14.2.2.4 Experimental results 14.2.2.4.1 Quantitative analysis 14.2.2.4.2 Parameter selection 14.2.2.4.3 Experiment results and analysis of the first three data sets 14.2.2.4.4 Experiment results and analysis on the last two image pairs 14.2.2.4.5 Results and analysis on simulated images 14.2.2.4.6 Experiment for sparse representation References 15. Hyperspectral image processing based on sparse learning and sparse graph 15.1 Hyperspectral image denoising based on hierarchical sparse learning 15.1.1 Spatial-spectral data extraction 15.1.2 Hierarchical sparse learning for denoising each band-subset 15.1.3 Experimental results and discussion 15.1.3.1 Experiment on simulated data 15.1.3.2 Experiment on real data 15.1.3.2.1 Denoising for urban data 15.1.3.2.2 Experimental results on Indian Pines data 15.2 Hyperspectral image restoration based on hierarchical sparse Bayesian learning 15.2.1 Beta process 15.2.1.1 Full hierarchical sparse Bayesian model 15.2.2 Experimental results 15.2.2.1 Denoising 15.2.2.2 Predicting the missing data 15.2.2.3 Discussion 15.3 Hyperspectral image dimensionality reduction using a sparse graph 15.3.1 Sparse representation 15.3.2 Sparse graph-based dimensionality reduction 15.3.3 Sparse graph learning 15.3.4 Spatial-spectral clustering 15.3.5 Experimental results 15.3.5.1 Introduction of hyperspectral datasets 15.3.5.2 Classification results 15.3.5.3 Influence of spatial-spectral clustering 15.3.5.4 Convergence analysis References 16. Nonconvex compressed sensing framework based on block strategy and overcomplete dictionary 16.1 Introduction 16.2 The block compressed sensing framework based on the overcomplete dictionary 16.2.1 Block compressed sensing 16.2.2 Overcomplete dictionary 16.2.3 Structured compressed sensing model 16.3 Image sparse representation based on the ridgelet overcomplete dictionary 16.4 Structured reconstruction model 16.4.1 Structural sparse prior based on image self-similarity 16.4.2 Reconstruction model based on an estimation of the direction structure of image blocks 16.5 Nonconvex reconstruction strategy References 17. Sparse representation combined with fuzzy C-means (FCM) in compressed sensing 17.1 Basic introduction to fuzzy C-means (FCM) and sparse representation (SR) 17.2 Two versions combining FCM with SR 17.2.1 FDCM_SSR 17.2.2 SL_FCM 17.3 Experimental results 17.3.1 FDCM_SSR 17.3.1.1 UCI data set 17.3.1.2 Artificial images 17.3.1.3 Natural images 17.4 SAR images 17.4.1 SL_FCM 17.4.1.1 Artificial and natural images 17.4.1.2 Synthetic aperture radar images References 18. Compressed sensing by collaborative reconstruction 18.1 Introduction 18.2 Methods 18.2.1 Block CS of images 18.2.2 Collaborative reconstruction method based on an overcomplete dictionary 18.2.3 Geometric structure-guided collaborative reconstruction method 18.3 Experiment 18.3.1 Collaborative reconstruction method based on an overcomplete dictionary 18.3.2 Geometric structure-guided collaborative reconstruction method References 19. Hyperspectral image classification based on spectral information divergence and sparse representation 19.1 The research status and challenges of hyperspectral image classification 19.1.1 The research status of hyperspectral image classification 19.1.2 The challenges of hyperspectral image classification 19.2 Motivation 19.3 Spectral information divergence (SID) 19.4 Sparse representation classification method based on SID 19.5 Joint sparse representation classification method based on SID 19.6 Experimental results and analysis 19.6.1 Comparison of the measurements 19.6.2 Comparison of the performance of sparse representation classification methods 19.6.3 Analysis of parameters 19.6.4 The proof of convergence References 20. Neural network-based synthetic aperture radar image processing 20.1 Discriminant deep belief network for SAR image classification 20.1.1 Weak classifiers training 20.1.2 Discriminative projection 20.1.3 High-level discriminative feature learning 20.1.4 Experiment and result 20.2 Convolutional-wavelet neural network for SAR image segmentation 20.2.1 Overall framework 20.2.2 Experiment and result 20.3 Deep neural network for SAR image registration 20.3.1 Train deep neural network 20.3.2 Predicting the matching label and eliminate the wrong matching points 20.3.3 Experiment and result References 21. Neural networks-based polarimetric SAR image classification 21.1 PolSAR decomposition 21.2 Autoencoder for PolSAR image classification 21.2.1 Data processing and feature learning 21.2.2 Experiment and result 21.3 DBN for PolSAR image classification 21.3.1 DBN structure and feature learning 21.3.2 Experiment and result 21.4 Wishart deep stacking networks for PolSAR image classification 21.4.1 Wishart distance and network structure 21.4.2 Experiment and results References 22. Deep neural network models for hyperspectral images 22.1 Deep fully convolutional network 22.1.1 Fully convolutional networks 22.1.2 Deep multiscale spatial distribution prediction via FCN-8s 22.1.3 Spatial-spectral feature fusion and classification for HSI 22.1.4 Experiment and results 22.2 Recursive autoencoders 22.2.1 Unsupervised RAE 22.2.2 Experiments and results 22.3 Superpixel-based multiple local CNN 22.3.1 Multiple local regions joint representation CNN model 22.3.2 Experiments and results References Index A B C D E F G H I J K L M N O P Q R S T U V W X Y Z 15.1 Hyperspectral image denoising based on hierarchical sparse learning 15.1.1 Spatial-spectral data extraction 15.1.2 Hierarchical sparse learning for denoising each band-subset 15.1.3 Experimental results and discussion 15.1.3.1 Experiment on simulated data 15.1.3.2 Experiment on real data 15.2 Hyperspectral image restoration based on hierarchical sparse Bayesian learning 15.2.1 Beta process 15.2.1.1 Full hierarchical sparse Bayesian model 15.2.2 Experimental results 15.2.2.1 Denoising 15.2.2.2 Predicting the missing data 15.2.2.3 Discussion 15.3 Hyperspectral image dimensionality reduction using a sparse graph 15.3.1 Sparse representation 15.3.2 Sparse graph-based dimensionality reduction 15.3.3 Sparse graph learning 15.3.4 Spatial-spectral clustering 15.3.5 Experimental results 15.3.5.1 Introduction of hyperspectral datasets 15.3.5.2 Classification results 15.3.5.3 Influence of spatial-spectral clustering 15.3.5.4 Convergence analysis References