
Artificial Neural Networks and Machine Learning - ICANN 2018
Description
Alles über E-Books | Antworten auf Fragen rund um E-Books, Kopierschutz und Dateiformate finden Sie in unserem Info- & Hilfebereich.
This three-volume set LNCS 11139-11141 constitutes the refereed proceedings of the 27 th International Conference on Artificial Neural Networks, ICANN 2018, held in Rhodes, Greece, in October 2018.
The papers presented in these volumes was carefully reviewed and selected from total of 360 submissions. They are related to the following thematic topics: AI and Bioinformatics, Bayesian and Echo State Networks, Brain Inspired Computing, Chaotic Complex Models, Clustering, Mining, Exploratory Analysis, Coding Architectures, Complex Firing Patterns, Convolutional Neural Networks, Deep Learning (DL), DL in Real Time Systems, DL and Big Data Analytics, DL and Big Data, DL and Forensics, DL and Cybersecurity, DL and Social Networks, Evolving Systems - Optimization, Extreme Learning Machines, From Neurons to Neuromorphism, From Sensation to Perception, From Single Neurons to Networks, Fuzzy Modeling, Hierarchical ANN, Inference and Recognition, Information and Optimization, Interacting with The Brain, Machine Learning (ML), ML for Bio Medical systems, ML and Video-Image Processing, ML and Forensics, ML and Cybersecurity, ML and Social Media, ML in Engineering, Movement and Motion Detection, Multilayer Perceptrons and Kernel Networks, Natural Language, Object and Face Recognition, Recurrent Neural Networks and Reservoir Computing, Reinforcement Learning, Reservoir Computing, Self-Organizing Maps, Spiking Dynamics/Spiking ANN, Support Vector Machines, Swarm Intelligence and Decision-Making, Text Mining, Theoretical Neural Computation, Time Series and Forecasting, Training and Learning.
More details
Other editions
Additional editions

Content
- Intro
- Preface
- Organization
- Keynote Talks
- Cognitive Phase Transitions in the Cerebral Cortex - John Taylor Memorial Lecture
- On the Deep Learning Revolution in Computer Vision
- From Machine Learning to Machine Diagnostics
- Multimodal Deep Learning in Biomedical Image Analysis
- Contents - Part III
- Recurrent ANN
- Policy Learning Using SPSA
- 1 Introduction
- 2 Simultaneous Perturbation Stochastic Approximation
- 3 Learning Policies Using Echo State Networks
- 3.1 Partial Observability
- 3.2 Echo State Networks
- 3.3 Policy Learning Using Echo State Networks
- 3.4 Deterministic and Stochastic Policies
- 3.5 Three Variants of Echo State Network Training
- 4 Experiments and Results
- 4.1 Acrobot and Mountain Car
- 4.2 Implementation Details
- 4.3 Results
- 5 Conclusion
- References
- Simple Recurrent Neural Networks for Support Vector Machine Training
- 1 Introduction
- 2 L2 Support Vector Machines
- 3 Frank-Wolfe Training of Support Vector Machines
- 4 Neural Training of Support Vector Machines
- 5 Practical Examples
- 6 Conclusion
- References
- RNN-SURV: A Deep Recurrent Model for Survival Analysis
- 1 Introduction
- 2 Related Work
- 3 Background on Survival Analysis
- 4 RNN-SURV
- 4.1 The Structure of the Model
- 4.2 Training
- 5 Experimental Analysis
- 5.1 Preprocessing
- 5.2 Comparison with Other Models
- 5.3 Estimating the Survival Curves
- 5.4 Analysis of the Model
- 6 Conclusions
- References
- Do Capsule Networks Solve the Problem of Rotation Invariance for Traffic Sign Classification?
- Abstract
- 1 Introduction
- 2 Related Work
- 3 Dataset
- 3.1 Preparation
- 3.2 Augmentation
- 4 Approach
- 5 Results
- 5.1 Accuracy
- 5.2 Rotation
- 6 Conclusion
- References
- Balanced and Deterministic Weight-Sharing Helps Network Performance
- 1 Introduction
- 1.1 ArbNet
- 1.2 How the Hash Function Affects Network Performance
- 2 Common Neural Networks are MLP ArbNets
- 2.1 Multi-layer Perceptrons
- 2.2 Convolutional Neural Networks
- 2.3 Recurrent Neural Networks
- 2.4 General Networks
- 3 Related Work
- 4 Experimental Setup
- 4.1 Balance of the Hash Table
- 4.2 Noise in the Hash Function
- 4.3 Network Specification
- 5 Results and Discussion
- 5.1 Dirichlet Hash
- 5.2 Neighborhood Hash
- 6 Conclusion
- References
- Neural Networks with Block Diagonal Inner Product Layers
- 1 Introduction
- 2 Related Work
- 3 Methodology
- 4 Experiments: Speedup and Accuracy
- 4.1 Speedup
- 4.2 Accuracy Results
- 5 Conclusion
- References
- Training Neural Networks Using Predictor-Corrector Gradient Descent
- 1 Introduction
- 2 Related Work
- 3 Methodology
- 4 Relationship to Nesterov's Accelerated Gradient
- 5 Experimental Results
- 5.1 SVHN
- 5.2 CIFAR10
- 6 Conclusion
- References
- Investigating the Role of Astrocyte Units in a Feedforward Neural Network
- 1 Introduction
- 2 Related Work
- 3 Proposed Models
- 3.1 A-MLP
- 3.2 A-MLP()
- 3.3 A-MLP()
- 3.4 A-MLP()
- 3.5 A-MLP(,), A-MLP(,,)
- 4 Experiments
- 4.1 N-parity Problem
- 4.2 Two-Spirals Problem
- 5 Conclusion
- References
- Interactive Area Topics Extraction with Policy Gradient
- 1 Introduction
- 2 Related Work
- 3 Methodology
- 3.1 Problem Formulation
- 3.2 LITE Model
- 4 Experimental Study
- 4.1 Experiment Configuration
- 4.2 Results and Discussion
- 5 Conclusion
- References
- Implementing Neural Turing Machines
- 1 Introduction
- 2 Neural Turing Machines
- 3 Our Implementation
- 4 Methodology
- 4.1 Tasks
- 4.2 Experiments
- 5 Results
- 5.1 Memory Initialization Comparison
- 5.2 Architecture Comparison
- 6 Summary
- References
- A RNN-Based Multi-factors Model for Repeat Consumption Prediction
- 1 Introduction
- 2 Related Works
- 3 Methodolody
- 3.1 RNN-based Multi-factors Prediction Model
- 3.2 Influential Factors Selection
- 4 Experiment
- 4.1 Dataset
- 4.2 Baselines Comparison
- 4.3 Influential Factor Analyze
- 5 Conclusion
- References
- Practical Fractional-Order Neuron Dynamics for Reservoir Computing
- 1 Introduction
- 2 Preliminaries
- 2.1 Reservoir Computing with Leaky Integrator Neurons
- 2.2 Learning of Readout Weights
- 3 Fractional-Order Neuron Dynamics
- 3.1 Derivation of Fractional-Order Leaky Integrator
- 3.2 Approximation to Memory Trace
- 4 Performance Evaluation
- 4.1 Benchmark Problems
- 4.2 Evaluation Criteria
- 4.3 Results
- 5 Conclusion
- References
- An Unsupervised Character-Aware Neural Approach to Word and Context Representation Learning
- 1 Introduction
- 2 Related Work
- 3 The Character-Aware Neural Model
- 3.1 Word and Context Embeddings
- 3.2 Learning Algorithm
- 4 Experimental Results
- 5 Conclusions
- References
- Towards End-to-End Raw Audio Music Synthesis
- 1 Introduction
- 2 Related Work
- 3 A Baseline Neural Model for Raw Audio Synthesis
- 4 Data Generation
- 5 Results and Evaluation
- 5.1 Empirical Evaluation
- 5.2 Qualitative Evaluation
- 6 Conclusion
- References
- Real-Time Hand Prosthesis Biomimetic Movement Based on Electromyography Sensory Signals Treatment and Sensors Fusion
- Abstract
- 1 Introduction
- 2 Methodology
- 2.1 Electromyography (EMG)
- 2.2 EMG Acquisition
- 2.3 Signal Treatment
- 2.4 The Prosthetic Hand
- 2.5 Database and Artificial Neural Networks
- 2.6 Sensors Fusion
- 3 Preliminary Results
- 4 Conclusion
- References
- An Exploration of Dropout with RNNs for Natural Language Inference
- 1 Introduction
- 2 Related Work
- 3 Recurrent Neural Network Model for NLI Task
- 4 Experiments and Results
- 4.1 Experimental Setup
- 4.2 Dropout at Different Layers for NLI Model
- 4.3 The Effectiveness of Dropout for Overfitting
- 4.4 Dropout Rate Effect on Accuracy and Dropout Location
- 5 Recommendations for Dropout Application
- 6 Conclusions
- References
- Neural Model for the Visual Recognition of Animacy and Social Interaction
- 1 Introduction
- 2 Stimulus Synthesis
- 3 Model Architecture
- 4 Results
- 5 Conclusion
- References
- Attention-Based RNN Model for Joint Extraction of Intent and Word Slot Based on a Tagging Strategy
- Abstract
- 1 Introduction
- 2 Related Works
- 3 Proposed Methods
- 3.1 The Tagging Strategy
- 3.2 Attention-Based RNN Model
- 4 Experiments
- 4.1 Experimental Settings
- 4.2 Intent Detection Task Results
- 4.3 Word Slot Extraction Task Results
- 4.4 Joint Task Results
- 5 Conclusion
- Acknowledgement
- References
- Using Regular Languages to Explore the Representational Capacity of Recurrent Neural Architectures
- 1 Introduction
- 2 Recurrent Neural Architectures for LDDs
- 3 Benchmarking Datasets
- 4 Formal Language Theory and Regular Languages
- 4.1 Strictly Piecewise Languages
- 5 Experiment
- 5.1 Generating SP2 dataset
- 5.2 Training Task
- 6 Analysis
- 7 Conclusion
- References
- Learning Trends on the Fly in Time Series Data Using Plastic CGP Evolved Recurrent Neural Networks
- Abstract
- 1 Introduction
- 2 Literature Review
- 3 Plastic Cartesian Genetic Programming Evolved Recurrent Neural Network
- 4 Experimental Setup
- 5 Results and Analysis
- 6 Conclusion and Future Enhancements
- References
- Noise Masking Recurrent Neural Network for Respiratory Sound Classification
- 1 Introduction
- 2 Related Work
- 3 Method
- 4 Experiments
- 4.1 Database
- 4.2 Experiments Setup
- 4.3 Result Evaluation
- 4.4 Preprocessing
- 5 Results
- 6 Conclusion
- References
- Lightweight Neural Programming: The GRPU
- 1 Introduction
- 2 Preliminaries
- 2.1 Gated Recurrent Unit (GRU)
- 2.2 Related Work: Neural Programmers
- 3 The Gated Recurrent Programmer Unit
- 3.1 The Architecture
- 3.2 The Arithmetic and Logic Unit (ALU)
- 3.3 Expanding the Model
- 4 Experimental Results
- 4.1 The Adding Problem
- 4.2 Other Variations
- 5 Conclusions and Future Work
- References
- Towards More Biologically Plausible Error-Driven Learning for Artificial Neural Networks
- References
- Online Carry Mode Detection for Mobile Devices with Compact RNNs
- 1 Introduction
- 2 Dataset
- 2.1 Acquisition
- 2.2 Dataset Preparation
- 3 Recurrent Neural Networks
- 3.1 Data Preprocessing
- 3.2 RNN Architecture
- 3.3 Training
- 3.4 Implementation Details
- 4 Experimental Results
- 4.1 Network Configurations
- 4.2 Results
- 5 Conclusion
- References
- Deep Learning
- Deep CNN-ELM Hybrid Models for Fire Detection in Images
- 1 Introduction
- 2 Related Work
- 2.1 CNNs for Fire Detection
- 2.2 Hybrid Models for Image Classification
- 3 The Fire Detector
- 3.1 Deep ConvNet Models
- 3.2 The Hybrid Model
- 3.3 Paper Contributions
- 4 Experiments
- 4.1 The Real World Fire Dataset
- 4.2 Results
- 5 Conclusion
- References
- Siamese Survival Analysis with Competing Risks
- 1 Introduction
- 1.1 Motivation
- 1.2 Related Works
- 1.3 Contributions
- 2 Problem Formulation
- 3 Siamese Survival Prognosis Network
- 4 Experiments
- 4.1 Hyper-Parameter Optimization
- 4.2 SEER
- 4.3 Synthetic Data
- 5 Conclusion
- References
- A Survey on Deep Transfer Learning
- 1 Introduction
- 2 Deep Transfer Learning
- 3 Categories
- 3.1 Instances-Based Deep Transfer Learning
- 3.2 Mapping-Based Deep Transfer Learning
- 3.3 Network-Based Deep Transfer Learning
- 3.4 Adversarial-Based Deep Transfer Learning
- 4 Conclusion
- References
- Cloud Detection in High-Resolution Multispectral Satellite Imagery Using Deep Learning
- 1 Introduction
- 2 Proposed Method
- 2.1 CloudPeru Dataset
- 2.2 Neural Network Training
- 3 Results
- 4 Conclusions
- References
- Metric Embedding Autoencoders for Unsupervised Cross-Dataset Transfer Learning
- 1 Introduction
- 2 Related Work
- 2.1 Deep Re-ID Methods
- 2.2 Deep Transfer Learning for Re-ID
- 3 Metric Embedding Learning
- 3.1 Loss Function
- 3.2 Network Architecture
- 3.3 Embedding Training
- 4 Unsupervised Transfer Learning of Embedding
- 4.1 Our Solution
- 4.2 Model Details
- 5 Experiments
- 5.1 Score Computation
- 5.2 Single Dataset Pre-training
- 5.3 Multiple Dataset Pre-training
- 5.4 Comparison with PUL
- 6 Conclusion
- References
- Classification of MRI Migraine Medical Data Using 3D Convolutional Neural Network
- 1 Introduction
- 2 Experimental Setup and Neural Network Architecture
- 2.1 Dataset Acquisition and Data Preprocessing
- 2.2 Network Architecture
- 2.3 Discriminative Regions Visualization
- 3 Experimental Results
- 3.1 Classification Result of CNN
- 3.2 Visualisation Result of CAM
- 4 Discussion, Conclusion and Future Work
- References
- Deep 3D Pose Dictionary: 3D Human Pose Estimation from Single RGB Image Using Deep Convolutional Neural Network
- Abstract
- 1 Introduction
- 2 Related Work
- 3 Proposed Approach
- 3.1 Building the 3D Pose Dictionary
- 3.2 Preprocessing and Augmentation
- 3.3 Training AlexNet for Classification
- 4 Experimental Results
- 4.1 Discussion
- 5 Conclusions
- Acknowledgments
- References
- FiLayer: A Novel Fine-Grained Layer-Wise Parallelism Strategy for Deep Neural Networks
- 1 Introduction
- 2 Related Works
- 3 Inter-layer Parallelism
- 3.1 Problem Analysis of Mini-batch Gradient Descent
- 3.2 Data Pipeline Algorithm
- 4 Intra-layer Parallelism
- 4.1 Analysis of Convolution Operation
- 4.2 Parallelization of Convolution Layer
- 5 Experimental Results
- 5.1 Datasets and Environments
- 5.2 Evaluating Inter-layer Parallelism
- 5.3 Evaluating Intra-layer Parallelism
- 6 Conclusions and Future Works
- References
- DeepVol: Deep Fruit Volume Estimation
- 1 Introduction
- 2 Related Work
- 3 DeepVol for Volume Estimation
- 3.1 Fruit Detection
- 3.2 Volume Estimation
- 4 Training Details
- 4.1 Loss Functions
- 4.2 Training Strategy
- 4.3 Optimizer and Regularization
- 5 Experimental Results
- 5.1 Dataset Collection
- 5.2 Performance Evaluation
- 6 Conclusion
- References
- Graph Matching and Pseudo-Label Guided Deep Unsupervised Domain Adaptation
- 1 Introduction
- 2 Proposed Approach
- 2.1 Problem Definition
- 2.2 Minimizing Domain Discrepancy with Graph Matching
- 2.3 Refinement with Pseudo-labels
- 3 Experiments and Results
- 4 Conclusions
- References
- fNIRS-Based Brain-Computer Interface Using Deep Neural Networks for Classifying the Mental State of Drivers
- 1 Introduction
- 2 Methods
- 2.1 Equipment
- 2.2 Data Capture
- 2.3 Pre-processing
- 2.4 Feature Extraction
- 2.5 Classifiers
- 3 Experimental Results
- 3.1 Comparison of DNN and RNN
- 3.2 Comparison with Common Classifiers
- 3.3 Changing the Inputs
- 4 Conclusions
- References
- Research on Fight the Landlords' Single Card Guessing Based on Deep Learning
- Abstract
- 1 Introduction
- 2 Fight the Landlords Games Introduction
- 3 Model Overall Frame Design
- 3.1 Data Cleaning
- 3.2 CNN Network Input and Output
- 3.3 The Use of CNN Model
- 3.4 The Use of CNN Model Evaluation Program
- 4 Model Overall Frame Design Experimental Results and Analysis
- 4.1 Single Card Prediction and Refinement Assessment Method
- 4.2 All Card Prediction
- 5 Conclusion
- Acknowledgment
- References
- Short-Term Precipitation Prediction with Skip-Connected PredNet
- Abstract
- 1 Introduction
- 2 Related Work
- 2.1 Video Generation with DNN
- 2.2 Network Architecture of DNN
- 3 Model
- 3.1 Problem Setting
- 3.2 Convolutional GRU
- 3.3 Our Model
- 3.4 Loss Functions
- 4 Experiments
- 4.1 MovingMNIST++
- 4.2 Kyoto Dataset
- 5 Conclusion
- Acknowledgment
- References
- An End-to-End Deep Learning Architecture for Classification of Malware's Binary Content
- 1 Introduction
- 2 Deep Learning for Malware Classification
- 3 Evaluation
- 3.1 Microsoft Malware Classification Challenge
- 3.2 Experimental Setup
- 3.3 State-of-the-art Comparison
- 4 Conclusions
- 4.1 Future Work
- References
- Width of Minima Reached by Stochastic Gradient Descent is Influenced by Learning Rate to Batch Size Ratio
- 1 Introduction
- 2 Theory
- 2.1 Learning Rate to Batch Size Ratio Determines SGD Dynamics
- 2.2 LR/BS Ratio Controls Trace of Hessian at a Minimum
- 3 Experiments
- 4 Related Work
- 5 Conclusion
- References
- Data Correction by a Generative Model with an Encoder and its Application to Structure Design
- 1 Introduction
- 2 Generative Models by Neural Networks
- 2.1 Generative Adversarial Nets
- 2.2 Wasserstein GAN
- 2.3 Variational Auto-Encoder
- 3 Proposed Model of GAN with an Encoder
- 3.1 Basic Model of GAN with an Encoder
- 3.2 Fine-Tuning with a Classifier on Data Space: Model 1
- 3.3 Fine-Tuning with a Classifier on Latent Space: Model 2
- 4 Computer Experiments Using Real Building Data
- 4.1 MNIST Handwritten Digits
- 4.2 Building Members Placement
- 5 Present Summary and Future Problems
- References
- PMGAN: Paralleled Mix-Generator Generative Adversarial Networks with Balance Control
- 1 Introduction
- 2 Related Works
- 3 Our Method
- 3.1 Loss Functions
- 3.2 The Balance Term
- 3.3 Structure of PMGAN
- 4 Experiments
- 4.1 Synthetic Datasets
- 4.2 Real World Data
- 4.3 Training Time
- 5 Conclusions
- References
- Modular Domain-to-Domain Translation Network
- Abstract
- 1 Introduction
- 2 The Proposed Model
- 2.1 Deep Domain to Domain Translation Architecture
- 3 Problem Setup
- 4 Experimental Results
- 5 Conclusions
- References
- OrieNet: A Regression System for Latent Fingerprint Orientation Field Extraction
- Abstract
- 1 Introduction
- 2 Proposed Method
- 2.1 Methods Overview
- 2.2 Latent Fingerprint Preprocessing
- 2.3 Deep Regression Neural Network
- 3 Experiments
- 3.1 Database
- 3.2 Identification Performance
- 4 Conclusion and Future Work
- Acknowledgement
- References
- Avoiding Degradation in Deep Feed-Forward Networks by Phasing Out Skip-Connections
- 1 Introduction
- 2 Related Work
- 3 Variable Activation Networks
- 4 Experiments
- 4.1 MNIST and Fashion-MNIST
- 4.2 CIFAR
- 5 Discussion
- References
- A Deep Predictive Coding Network for Inferring Hierarchical Causes Underlying Sensory Inputs
- Abstract
- 1 Introductions
- 2 Model
- 2.1 Architecture
- 2.2 Learning Algorithm
- 3 Experiments
- 3.1 Generative Model
- 3.2 Capacity to Represent Novel Input Patterns
- 3.3 Robustness Towards Translated Images
- 4 Discussion
- 5 Conclusion
- Acknowledgement
- References
- Type-2 Diabetes Mellitus Diagnosis from Time Series Clinical Data Using Deep Learning Models
- 1 Introduction
- 2 Related Work
- 3 Dataset
- 3.1 Data Pre-processing
- 4 Methods
- 5 Experimental Setup
- 6 Results
- 6.1 Discussion and Conclusion
- References
- A Deep Learning Approach for Sentence Classification of Scientific Abstracts
- 1 Introduction
- 2 Related Work
- 3 Materials and Methods
- 3.1 Abstract Corpus
- 3.2 Neural Networks Models
- 3.3 Proposed Architecture
- 3.4 Evaluation
- 4 Results
- 5 Conclusions
- References
- Weighted Multi-view Deep Neural Networks for Weather Forecasting
- 1 Introduction
- 2 Background: Multi-View LS-SVM Regression
- 3 Proposed Method
- 4 Experiments
- 4.1 Weather Data
- 4.2 Model Selection
- 5 Conclusion
- References
- Combining Articulatory Features with End-to-End Learning in Speech Recognition
- Abstract
- 1 Introduction
- 2 Related Work
- 2.1 End-to-End Learning in Speech Recognition
- 2.2 Domain Knowledge Integration in Speech Recognition
- 3 Model Architecture
- 3.1 AF Extractor
- 3.2 Fine-Tuning Networks
- 3.3 Progressive Networks
- 4 Experiments
- 4.1 Evaluation Metric
- 4.2 Data
- 4.3 Training
- 5 Results and Discussion
- 6 Conclusions and Future Work
- Acknowledgements
- References
- Estimation of Air Quality Index from Seasonal Trends Using Deep Neural Network
- 1 Introduction
- 2 Related Previous Work
- 3 Motivation and Problem Statement
- 3.1 Motivation
- 3.2 Problem Statement
- 4 Proposed Method for Estimation of AQI
- 4.1 Formation of Sub-indices (Step 1)
- 4.2 Aggregation of Sub-indices (Step 2)
- 4.3 Proposed Models
- 4.4 ARIMA Model
- 4.5 RNN-Based Model
- 4.6 Seasonal Data Re-configuring
- 4.7 Architecture of RNN-Layer
- 5 Simulation Results
- 5.1 Input Dataset
- 5.2 Comparative Results and Discussions
- 6 Conclusions
- References
- A Deep Learning Approach to Bacterial Colony Segmentation
- 1 Introduction
- 1.1 Related Works
- 2 Synthetic Petri Plate Generation
- 2.1 Background and Token Collection
- 2.2 Colony Models
- 2.3 Streaking Simulation
- 2.4 Rendering and Blending Procedure
- 3 Experiments
- 3.1 Semantic Segmentation Network
- 3.2 Dataset
- 3.3 Segmentation Experiments
- 4 Conclusions
- References
- Sparsity and Complexity of Networks Computing Highly-Varying Functions
- 1 Introduction
- 2 Preliminaries
- 3 Sparsity, Variational Norm, and Correlation
- 4 Lower Bounds on l1 and Variational Norms
- 5 Construction of Functions with Large Variations with Respect to Perceptrons
- 6 Discussion
- References
- Deep Learning Based Vehicle Make-Model Classification
- 1 Introduction
- 2 Vehicle Make-Model Classification
- 2.1 Data Gathering
- 2.2 Annotation
- 2.3 Model Training and Testing
- 3 Experimental Results
- 3.1 A Use Case
- 4 Conclusions and Future Works
- References
- Detection and Recognition of Badgers Using Deep Learning
- Abstract
- 1 Introduction
- 2 Dataset and Preprocessing
- 3 Methods
- 3.1 SSD with Inception-V2
- 3.2 SSD with MobileNet-V1
- 3.3 Faster R-CNN with ResNet-50
- 3.4 Faster R-CNN with Inception-V2
- 4 Experimental Results
- 5 Conclusion
- References
- SPSA for Layer-Wise Training of Deep Networks
- 1 Introduction
- 2 Derivative-Free Optimization and SPSA
- 3 SPSA-Based Neural Network Training
- 4 Practical Experiments
- 4.1 Low-Dimensional Classification Problems
- 4.2 MNIST
- 5 Conclusion
- References
- Dipolar Data Aggregation in the Context of Deep Learning
- Abstract
- 1 Introduction
- 2 Partially Structured Data Sets
- 3 Separable Layers of Univariate Binary Classifiers
- 4 Separation of Selected Data Subsets by Dipolar Layers of Univariate Binary Classifiers
- 5 Hierarchical Networks of Separable Layers
- 6 Experimental Results
- 7 Concluding Remarks
- Acknowledgments
- References
- Video Surveillance of Highway Traffic Events by Deep Learning Architectures
- 1 Introduction
- 2 Video Surveillance of Highway Traffic
- 3 Data Description and Representation
- 4 Deep Architectures
- 5 Experimental Results
- 6 Conclusions
- References
- Augmenting Image Classifiers Using Data Augmentation Generative Adversarial Networks
- 1 Introduction
- 2 Background
- 3 Model
- 3.1 Model Overview
- 3.2 Model Objective Definition
- 3.3 Architectures
- 4 Datasets and Experiments
- 4.1 Training of DAGAN in Source Domain
- 4.2 Classifiers
- 5 Conclusions
- References
- DeepEthnic: Multi-label Ethnic Classification from Face Images
- 1 Introduction
- 2 Related Work
- 2.1 Traditional ML-Based Techniques
- 2.2 Recent Deep Learning Techniques
- 3 Proposed Method
- 3.1 Data Source
- 3.2 Facial Image Preprocessing
- 3.3 Transfer Learning
- 3.4 Network Architecture
- 4 Experiments and Results
- 5 Conclusions
- References
- Handwriting-Based Gender Classification Using End-to-End Deep Neural Networks
- 1 Introduction
- 2 Related Work
- 3 Proposed Method
- 3.1 The HEBIU Offline Handwriting Dataset
- 3.2 Handwriting Preprocessing
- 3.3 Network Architecture
- 3.4 Accuracy Evaluation by Patch Aggregation
- 4 Experimental Results
- 4.1 Intra-language Classification
- 4.2 Inter-language Classification
- 4.3 Mixed Language Classification
- 4.4 Summary of Results
- 4.5 Human Test Results
- 5 Concluding Remarks
- References
- A Deep Learning Approach for Sentiment Analysis in Spanish Tweets
- Abstract
- 1 Introduction
- 2 Related Studies
- 3 The Problem and Data Description
- 4 Methodology
- 4.1 Preprocessing
- 4.2 Word Vectors
- 4.3 Convolutional Neural Network
- 5 Experiments and Results
- 5.1 Filters Setting
- 5.2 Sentiment Analysis
- 6 Conclusions and Future Works
- References
- Location Dependency in Video Prediction
- 1 Introduction
- 2 Related Work
- 3 Location Dependency in VLN Model
- 4 Location Dependency in Conv-PGP Model
- 5 Experiment
- 6 Conclusion
- References
- Brain Neurocomputing Modeling
- State-Space Analysis of an Ising Model Reveals Contributions of Pairwise Interactions to Sparseness, Fluctuation, and Stimulus Coding of Monkey V1 Neurons
- 1 Introduction
- 2 Methods
- 2.1 Data Description and Preprocessing
- 2.2 The State-Space Ising Model for a Neural Population
- 2.3 Macroscopic Properties of the Dynamic Ising Model
- 2.4 Assessment of Stimulus Coding
- 3 Results
- 3.1 Contributions of Interactions to Macroscopic Network Properties
- 3.2 Differences in Neural Responses Caused by Different Stimuli
- 4 Discussion
- 5 Conclusion
- References
- Sparse Coding Predicts Optic Flow Specifities of Zebrafish Pretectal Neurons
- 1 Introduction
- 2 Visual Front End
- 3 LCA Sparse Coding
- 4 Backpropagation
- 5 Results
- 6 Discussion
- References
- Brain-Machine Interface for Mechanical Ventilation Using Respiratory-Related Evoked Potential
- 1 Introduction
- 2 Methods
- 2.1 Existing Approaches
- 2.2 Riemannian Geometry
- 3 Experiments
- 3.1 Setup and Dataset
- 3.2 Results
- 4 Discussion and Conclusion
- References
- Effectively Interpreting Electroencephalogram Classification Using the Shapley Sampling Value to Prune a Feature Tree
- Abstract
- 1 Introduction
- 2 Method
- 3 Experimental Settings
- 3.1 PhysioNet Polysomnography Dataset
- 3.2 UCI EEG Dataset
- 4 Results
- 4.1 Results for the PhysioNet PSG Dataset
- 4.2 Results for the UCI EEG Dataset
- 5 Discussion and Conclusion
- Acknowledgements
- References
- EEG-Based Person Identification Using Rhythmic Brain Activity During Sleep
- Abstract
- 1 Introduction
- 2 Method
- 2.1 Feature Extraction
- 3 Experiments
- 3.1 The Sleep Spindles Database
- 3.2 Data Preprocessing
- 3.3 Experimental Setup
- 4 Results
- 4.1 Person Identification Using All Channels
- 4.2 Person Identification Using Single Channel
- 4.3 Person Identification Using Significant Channels
- 4.4 Person Identification Using Significant Features
- 5 Conclusions
- References
- An STDP Rule for the Improvement and Stabilization of the Attractor Dynamics of the Basal Ganglia-Thalamocortical Network
- 1 Introduction
- 2 Boolean Model of the Basal Ganglia-Thalamocortical Network
- 3 Adaptive STDP Rule
- 4 Results
- 5 Conclusion
- References
- Neuronal Asymmetries and Fokker-Planck Dynamics
- 1 Introduction
- 2 Continuous Neural Networks and Dynamical Systems
- 3 Standard Fokker-Planck Dynamics
- 4 Generalized Fokker-Planck Dynamics
- 4.1 Stationary Solution for Drift Fields of Gradient Form
- 4.2 Stationary Solution for K with Non-Gradient Components
- 5 A Two-Neuron System Admitting Time-Dependent q-Gaussian Solutions
- 6 Conclusions
- References
- Robotics/Motion Detection
- Learning-While Controlling RBF-NN for Robot Dynamics Approximation in Neuro-Inspired Control of Switched Nonlinear Systems
- Abstract
- 1 Introduction
- 2 Methodology
- 3 Controller
- 4 Results
- 5 Discussion and Conclusion
- Acknowledgments
- References
- A Feedback Neural Network for Small Target Motion Detection in Cluttered Backgrounds
- 1 Introduction
- 2 Formulation of the Model
- 2.1 Retina Layer
- 2.2 Lamina Layer
- 2.3 Medulla Layer
- 2.4 Lobula Layer
- 3 Results and Discussions
- 4 Conclusion
- References
- De-noise-GAN: De-noising Images to Improve RoboCup Soccer Ball Detection
- 1 Introduction
- 2 De-noising Generative Adversarial Network
- 2.1 Generative Models
- 2.2 DCGANs
- 2.3 De-noise-GAN
- 3 Experimental Results
- 3.1 Dataset and Acquisition
- 3.2 De-noising
- 3.3 Ball Localization
- 4 Discussion
- 5 Conclusion
- 6 Future Work
- References
- Integrative Collision Avoidance Within RNN-Driven Many-Joint Robot Arms
- 1 Introduction
- 2 Method
- 2.1 Robot Arm Model and Selective Control
- 2.2 Local Distance Sensor Signals
- 2.3 Sensory Gradient Injection
- 3 Experiments
- 3.1 Moving Box
- 3.2 Many Boxes
- 3.3 Wall Opening
- 4 Summary and Conclusion
- References
- An Improved Block-Matching Algorithm Based on Chaotic Sine-Cosine Algorithm for Motion Estimation
- 1 Introduction
- 2 Basic Preliminaries
- 2.1 Motion Estimation and Block-Matching
- 2.2 Sine-Cosine Algorithm (SCA)
- 2.3 Chaotic Maps
- 3 Proposed Chaotic-Based SCA with Fitness Approximation Strategy
- 4 Discussion and Analysis of the Results
- 4.1 Detailed Analysis with Respect to PSNR and DPSNR
- 4.2 Detailed Analysis with Respect to the Number of Search Counts
- 5 Conclusion
- References
- Terrain Classification with Crawling Robot Using Long Short-Term Memory Network
- Abstract
- 1 Introduction
- 2 Proprioceptive Signals and Data Collection
- 2.1 Adaptive Gait
- 2.2 Data Collection and Preprocessing
- 3 Proposed Terrain Predictor
- 4 Experiments
- 5 Conclusion
- Acknowledgments
- References
- Mass-Spring Damper Array as a Mechanical Medium for Computation
- Abstract
- 1 Introduction
- 2 Mass-Spring Damper Array
- 3 Approximation of NARMA Models as a Benchmark Test
- 4 Pretests with Echo State Networks
- 5 Results of the Benchmark Tests of the Mass-Spring Damper Array
- 6 Discussion
- 7 Conclusion
- References
- Kinematic Estimation with Neural Networks for Robotic Manipulators
- 1 Introduction
- 2 Related Work
- 3 Problem Formulation
- 4 Experimental Testbed
- 4.1 Kinematics of the Sawyer Robot
- 4.2 Network Architecture
- 4.3 Experimental Results
- 5 Conclusions
- References
- Social Media
- Hierarchical Attention Networks for User Profile Inference in Social Media Systems
- 1 Introduction
- 2 Related Work
- 3 Hierarchical Attention Networks
- 3.1 GRU-Based Sequence Encoder
- 3.2 Hierarchical Attention Based GRU Neural Network
- 3.3 Hierarchical Inner Attention Based Neural Network
- 3.4 Ego Network Classification
- 4 Experiments
- 4.1 Datasets
- 4.2 User Representation and Dimension Selection
- 4.3 Comparative Methods and Experiment Setting
- 4.4 Result and Analysis
- 5 Conclusion
- References
- A Topological k-Anonymity Model Based on Collaborative Multi-view Clustering
- 1 Introduction
- 2 Related Works
- 2.1 Clustering and Anonymization
- 2.2 Multi-view Topological Collaborative Clustering
- 3 Proposed Anonymization Model
- 3.1 Experimental Protocol
- 3.2 Data Sets
- 3.3 Experimental Results
- 4 Conclusion
- References
- A Credibility-Based Analysis of Information Diffusion in Social Networks
- Abstract
- 1 Introduction
- 1.1 Dempster-Shafer Theory
- 1.2 Yager's Rule
- 2 Related Work
- 3 Model Description
- 3.1 The Information Diffusion Protocol
- 3.2 Computing Confidence Degrees
- 4 Simulation Results
- 5 Conclusions
- References
- Author Index
System requirements
File format: PDF
Copy protection: Watermark-DRM (Digital Rights Management)
System requirements:
- Computer (Windows; MacOS X; Linux): Use the free software Adobe Reader, Adobe Digital Editions, or any other PDF viewer of your choice (see eBook Help).
- Tablet/Smartphone (Android; iOS): Install the free app Adobe Digital Editions or another reading app for eBooks, e.g., PocketBook (see eBook Help).
- E-reader: Bookeen, Kobo, Pocketbook, Sony, Tolino and many more (only limited: Kindle).
The file format PDF always displays a book page identically on any hardware. This makes PDF suitable for complex layouts such as those used in textbooks and reference books (images, tables, columns, footnotes). Unfortunately, on the small screens of e-readers or smartphones, PDFs are rather annoying, requiring too much scrolling.
This eBook uses Watermark-DRM, a „soft” copy protection. This means that there are no technical restrictions to prevent illegal distribution. However, there is a personalised watermark embedded in the eBook that can be used to identify the purchaser of the eBook in the event of misuse and to provide evidence for legal purposes.
For more information, see our eBook Help page.